GNAT Reference Manual
GNAT, The GNU Ada 95 Compiler
GNAT Version for GCC 3.3.5
Ada Core Technologies, Inc.
Copyright © 1995-2001, Free Software Foundation
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with the Invariant Sections being “GNU Free Documentation License”, with the Front-Cover Texts being “GNAT Reference Manual”, and with no Back-Cover Texts. A copy of the license is included in the section entitled “GNU Free Documentation License”.
--- The Detailed Node Listing ---
About This Guide
The Implementation of Standard I/O
The GNAT Library
Text_IO
Wide_Text_IO
Interfacing to Other Languages
GNAT Implementation of Tasking
This manual contains useful information in writing programs using the GNAT compiler. It includes information on implementation dependent characteristics of GNAT, including all the information required by Annex M of the standard.
Ada 95 is designed to be highly portable,and guarantees that, for most programs, Ada 95 compilers behave in exactly the same manner on different machines. However, since Ada 95 is designed to be used in a wide variety of applications, it also contains a number of system dependent features to Functbe used in interfacing to the external world.
Note: Any program that makes use of implementation-dependent features may be non-portable. You should follow good programming practice and isolate and clearly document any sections of your program that make use of these features in a non-portable manner.
This reference manual contains the following chapters:
This reference manual assumes that you are familiar with Ada 95 language, as described in the International Standard ANSI/ISO/IEC-8652:1995, Jan 1995.
Following are examples of the typographical and graphic conventions used in this guide:
Functions, utility program names, standard names,
and classes.
Option flags
Variables.
and then shown this way.
Commands that are entered by the user are preceded in this manual by the characters $ (dollar sign followed by space). If your system uses this sequence as a prompt, then the commands will appear exactly as you see them in the manual. If your system uses some other prompt, then the command will appear with the $ replaced by whatever prompt character you are using.
See the following documents for further information on GNAT:
Ada 95 defines a set of pragmas that can be used to supply additional information to the compiler. These language defined pragmas are implemented in GNAT and work as described in the Ada 95 Reference Manual.
In addition, Ada 95 allows implementations to define additional pragmas whose meaning is defined by the implementation. GNAT provides a number of these implementation-dependent pragmas which can be used to extend and enhance the functionality of the compiler. This section of the GNAT Reference Manual describes these additional pragmas.
Note that any program using these pragmas may not be portable to other compilers (although GNAT implements this set of pragmas on all platforms). Therefore if portability to other compilers is an important consideration, the use of these pragmas should be minimized.
pragma Abort_Defer pragma Abort_Defer;
This pragma must appear at the start of the statement sequence of a
handled sequence of statements (right after the begin). It has
the effect of deferring aborts for the sequence of statements (but not
for the declarations or handlers, if any, associated with this statement
sequence).
pragma Ada_83 pragma Ada_83;
A configuration pragma that establishes Ada 83 mode for the unit to
which it applies, regardless of the mode set by the command line
switches. In Ada 83 mode, GNAT attempts to be as compatible with
the syntax and semantics of Ada 83, as defined in the original Ada
83 Reference Manual as possible. In particular, the new Ada 95
keywords are not recognized, optional package bodies are allowed,
and generics may name types with unknown discriminants without using
the (<>) notation. In addition, some but not all of the additional
restrictions of Ada 83 are enforced.
Ada 83 mode is intended for two purposes. Firstly, it allows existing legacy Ada 83 code to be compiled and adapted to GNAT with less effort. Secondly, it aids in keeping code backwards compatible with Ada 83. However, there is no guarantee that code that is processed correctly by GNAT in Ada 83 mode will in fact compile and execute with an Ada 83 compiler, since GNAT does not enforce all the additional checks required by Ada 83.
pragma Ada_95 pragma Ada_95;
A configuration pragma that establishes Ada 95 mode for the unit to which
it applies, regardless of the mode set by the command line switches.
This mode is set automatically for the Ada and System
packages and their children, so you need not specify it in these
contexts. This pragma is useful when writing a reusable component that
itself uses Ada 95 features, but which is intended to be usable from
either Ada 83 or Ada 95 programs.
pragma Annotate pragma Annotate (IDENTIFIER {, ARG});
ARG ::= NAME | EXPRESSION
This pragma is used to annotate programs. identifier identifies
the type of annotation. GNAT verifies this is an identifier, but does
not otherwise analyze it. The arg argument
can be either a string literal or an
expression. String literals are assumed to be of type
Standard.String. Names of entities are simply analyzed as entity
names. All other expressions are analyzed as expressions, and must be
unambiguous.
The analyzed pragma is retained in the tree, but not otherwise processed by any part of the GNAT compiler. This pragma is intended for use by external tools, including ASIS.
pragma Assert pragma Assert (
boolean_EXPRESSION
[, static_string_EXPRESSION])
The effect of this pragma depends on whether the corresponding command line switch is set to activate assertions. The pragma expands into code equivalent to the following:
if assertions-enabled then
if not boolean_EXPRESSION then
System.Assertions.Raise_Assert_Failure
(string_EXPRESSION);
end if;
end if;
The string argument, if given, is the message that will be associated with the exception occurrence if the exception is raised. If no second argument is given, the default message is file:nnn, where file is the name of the source file containing the assert, and nnn is the line number of the assert. A pragma is not a statement, so if a statement sequence contains nothing but a pragma assert, then a null statement is required in addition, as in:
...
if J > 3 then
pragma Assert (K > 3, "Bad value for K");
null;
end if;
Note that, as with the if statement to which it is equivalent, the
type of the expression is either Standard.Boolean, or any type derived
from this standard type.
If assertions are disabled (switch -gnata not used), then there
is no effect (and in particular, any side effects from the expression
are suppressed). More precisely it is not quite true that the pragma
has no effect, since the expression is analyzed, and may cause types
to be frozen if they are mentioned here for the first time.
If assertions are enabled, then the given expression is tested, and if
it is False then System.Assertions.Raise_Assert_Failure is called
which results in the raising of Assert_Failure with the given message.
If the boolean expression has side effects, these side effects will turn on and off with the setting of the assertions mode, resulting in assertions that have an effect on the program. You should generally avoid side effects in the expression arguments of this pragma. However, the expressions are analyzed for semantic correctness whether or not assertions are enabled, so turning assertions on and off cannot affect the legality of a program.
pragma Ast_Entry pragma AST_Entry (entry_IDENTIFIER);
This pragma is implemented only in the OpenVMS implementation of GNAT. The
argument is the simple name of a single entry; at most one AST_Entry
pragma is allowed for any given entry. This pragma must be used in
conjunction with the AST_Entry attribute, and is only allowed after
the entry declaration and in the same task type specification or single task
as the entry to which it applies. This pragma specifies that the given entry
may be used to handle an OpenVMS asynchronous system trap (AST)
resulting from an OpenVMS system service call. The pragma does not affect
normal use of the entry. For further details on this pragma, see the
DEC Ada Language Reference Manual, section 9.12a.
pragma C_Pass_By_Copy pragma C_Pass_By_Copy
([Max_Size =>] static_integer_EXPRESSION);
Normally the default mechanism for passing C convention records to C
convention subprograms is to pass them by reference, as suggested by RM
B.3(69). Use the configuration pragma C_Pass_By_Copy to change
this default, by requiring that record formal parameters be passed by
copy if all of the following conditions are met:
Convention C.
If these conditions are met the argument is passed by copy, i.e. in a manner consistent with what C expects if the corresponding formal in the C prototype is a struct (rather than a pointer to a struct).
You can also pass records by copy by specifying the convention
C_Pass_By_Copy for the record type, or by using the extended
Import and Export pragmas, which allow specification of
passing mechanisms on a parameter by parameter basis.
pragma Comment pragma Comment (static_string_EXPRESSION);
This is almost identical in effect to pragma Ident. It allows the
placement of a comment into the object file and hence into the
executable file if the operating system permits such usage. The
difference is that Comment, unlike Ident, has no limit on the
length of the string argument, and no limitations on placement
of the pragma (it can be placed anywhere in the main source unit).
pragma Common_Object pragma Common_Object (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Size =>] EXTERNAL_SYMBOL] )
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
This pragma enables the shared use of variables stored in overlaid
linker areas corresponding to the use of COMMON
in Fortran. The single
object local_name is assigned to the area designated by
the External argument.
You may define a record to correspond to a series
of fields. The size argument
is syntax checked in GNAT, but otherwise ignored.
Common_Object is not supported on all platforms. If no
support is available, then the code generator will issue a message
indicating that the necessary attribute for implementation of this
pragma is not available.
pragma Complex_Representation pragma Complex_Representation
([Entity =>] LOCAL_NAME);
The Entity argument must be the name of a record type which has two fields of the same floating-point type. The effect of this pragma is to force gcc to use the special internal complex representation form for this record, which may be more efficient. Note that this may result in the code for this type not conforming to standard ABI (application binary interface) requirements for the handling of record types. For example, in some environments, there is a requirement for passing records by pointer, and the use of this pragma may result in passing this type in floating-point registers.
pragma Component_Alignment pragma Component_Alignment (
[Form =>] ALIGNMENT_CHOICE
[, [Name =>] type_LOCAL_NAME]);
ALIGNMENT_CHOICE ::=
Component_Size
| Component_Size_4
| Storage_Unit
| Default
Specifies the alignment of components in array or record types. The meaning of the Form argument is as follows:
Component_SizeComponent_Size_4Storage_UnitSystem.Storage_Unit.
DefaultDefault choice is the same as
the Storage_Unit choice (byte alignment). For all other systems,
the Default choice is the same as Component_Size (natural
alignment).
If the Name parameter is present, type_local_name must
refer to a local record or array type, and the specified alignment
choice applies to the specified type. The use of
Component_Alignment together with a pragma Pack causes the
Component_Alignment pragma to be ignored. The use of
Component_Alignment together with a record representation clause
is only effective for fields not specified by the representation clause.
If the Name parameter is absent, the pragma can be used as either
a configuration pragma, in which case it applies to one or more units in
accordance with the normal rules for configuration pragmas, or it can be
used within a declarative part, in which case it applies to types that
are declared within this declarative part, or within any nested scope
within this declarative part. In either case it specifies the alignment
to be applied to any record or array type which has otherwise standard
representation.
If the alignment for a record or array type is not specified (using
pragma Pack, pragma Component_Alignment, or a record rep
clause), the GNAT uses the default alignment as described previously.
pragma Convention_Identifier pragma Convention_Identifier (
[Name =>] IDENTIFIER,
[Convention =>] convention_IDENTIFIER);
This pragma provides a mechanism for supplying synonyms for existing
convention identifiers. The Name identifier can subsequently
be used as a synonym for the given convention in other pragmas (including
for example pragma Import or another Convention_Identifier
pragma). As an example of the use of this, suppose you had legacy code
which used Fortran77 as the identifier for Fortran. Then the pragma:
pragma Convention_Indentifier (Fortran77, Fortran);
would allow the use of the convention identifier Fortran77 in
subsequent code, avoiding the need to modify the sources. As another
example, you could use this to parametrize convention requirements
according to systems. Suppose you needed to use Stdcall on
windows systems, and C on some other system, then you could
define a convention identifier Library and use a single
Convention_Identifier pragma to specify which convention
would be used system-wide.
pragma CPP_Class pragma CPP_Class ([Entity =>] LOCAL_NAME);
The argument denotes an entity in the current declarative region that is declared as a tagged or untagged record type. It indicates that the type corresponds to an externally declared C++ class type, and is to be laid out the same way that C++ would lay out the type.
If (and only if) the type is tagged, at least one component in the
record must be of type Interfaces.CPP.Vtable_Ptr, corresponding
to the C++ Vtable (or Vtables in the case of multiple inheritance) used
for dispatching.
Types for which CPP_Class is specified do not have assignment or
equality operators defined (such operations can be imported or declared
as subprograms as required). Initialization is allowed only by
constructor functions (see pragma CPP_Constructor).
Pragma CPP_Class is intended primarily for automatic generation
using an automatic binding generator tool.
See Interfacing to C++ for related information.
pragma CPP_Constructor pragma CPP_Constructor ([Entity =>] LOCAL_NAME);
This pragma identifies an imported function (imported in the usual way
with pragma Import) as corresponding to a C++
constructor. The argument is a name that must have been
previously mentioned in a pragma Import
with Convention = CPP, and must be of one of the following
forms:
function Fname return T'Class
function Fname (...) return T'Class
where T is a tagged type to which the pragma CPP_Class applies.
The first form is the default constructor, used when an object of type T is created on the Ada side with no explicit constructor. Other constructors (including the copy constructor, which is simply a special case of the second form in which the one and only argument is of type T), can only appear in two contexts:
Although the constructor is described as a function that returns a value on the Ada side, it is typically a procedure with an extra implicit argument (the object being initialized) at the implementation level. GNAT issues the appropriate call, whatever it is, to get the object properly initialized.
In the case of derived objects, you may use one of two possible forms for declaring and creating an object:
New_Object : Derived_T
New_Object : Derived_T := (constructor-function-call with ...)
In the first case the default constructor is called and extension fields if any are initialized according to the default initialization expressions in the Ada declaration. In the second case, the given constructor is called and the extension aggregate indicates the explicit values of the extension fields.
If no constructors are imported, it is impossible to create any objects on the Ada side. If no default constructor is imported, only the initialization forms using an explicit call to a constructor are permitted.
Pragma CPP_Constructor is intended primarily for automatic generation
using an automatic binding generator tool.
See Interfacing to C++ for more related information.
pragma CPP_Virtual pragma CPP_Virtual
[Entity =>] ENTITY,
[, [Vtable_Ptr =>] vtable_ENTITY,]
[, [Position =>] static_integer_EXPRESSION])
This pragma serves the same function as pragma Import in that
case of a virtual function imported from C++. The Entity argument
must be a
primitive subprogram of a tagged type to which pragma CPP_Class
applies. The Vtable_Ptr argument specifies
the Vtable_Ptr component which contains the
entry for this virtual function. The Position argument
is the sequential number
counting virtual functions for this Vtable starting at 1.
The Vtable_Ptr and Position arguments may be omitted if
there is one Vtable_Ptr present (single inheritance case) and all
virtual functions are imported. In that case the compiler can deduce both
these values.
No External_Name or Link_Name arguments are required for a
virtual function, since it is always accessed indirectly via the
appropriate Vtable entry.
Pragma CPP_Virtual is intended primarily for automatic generation
using an automatic binding generator tool.
See Interfacing to C++ for related information.
pragma CPP_Vtable pragma CPP_Vtable (
[Entity =>] ENTITY,
[Vtable_Ptr =>] vtable_ENTITY,
[Entry_Count =>] static_integer_EXPRESSION);
Given a record to which the pragma CPP_Class applies,
this pragma can be specified for each component of type
CPP.Interfaces.Vtable_Ptr.
Entity is the tagged type, Vtable_Ptr
is the record field of type Vtable_Ptr, and Entry_Count is
the number of virtual functions on the C++ side. Not all of these
functions need to be imported on the Ada side.
You may omit the CPP_Vtable pragma if there is only one
Vtable_Ptr component in the record and all virtual functions are
imported on the Ada side (the default value for the entry count in this
case is simply the total number of virtual functions).
Pragma CPP_Vtable is intended primarily for automatic generation
using an automatic binding generator tool.
See Interfacing to C++ for related information.
pragma Debug pragma Debug (PROCEDURE_CALL_WITHOUT_SEMICOLON);
PROCEDURE_CALL_WITHOUT_SEMICOLON ::=
PROCEDURE_NAME
| PROCEDURE_PREFIX ACTUAL_PARAMETER_PART
The argument has the syntactic form of an expression, meeting the syntactic requirements for pragmas.
If assertions are not enabled on the command line, this pragma has no
effect. If asserts are enabled, the semantics of the pragma is exactly
equivalent to the procedure call statement corresponding to the argument
with a terminating semicolon. Pragmas are permitted in sequences of
declarations, so you can use pragma Debug to intersperse calls to
debug procedures in the middle of declarations.
pragma Elaboration_Checks pragma Elaboration_Checks (RM | Static);
This is a configuration pragma that provides control over the
elaboration model used by the compilation affected by the
pragma. If the parameter is RM, then the dynamic elaboration
model described in the Ada Reference Manual is used, as though
the -gnatE switch had been specified on the command
line. If the parameter is Static, then the default GNAT static
model is used. This configuration pragma overrides the setting
of the command line. For full details on the elaboration models
used by the GNAT compiler, see section “Elaboration Order
Handling in GNAT” in the GNAT User's Guide.
pragma Eliminate pragma Eliminate (
[Unit_Name =>] IDENTIFIER |
SELECTED_COMPONENT);
pragma Eliminate (
[Unit_Name =>] IDENTIFIER |
SELECTED_COMPONENT,
[Entity =>] IDENTIFIER |
SELECTED_COMPONENT |
STRING_LITERAL
[,[Parameter_Types =>] PARAMETER_TYPES]
[,[Result_Type =>] result_SUBTYPE_NAME]
[,[Homonym_Number =>] INTEGER_LITERAL]);
PARAMETER_TYPES ::= (SUBTYPE_NAME {, SUBTYPE_NAME})
SUBTYPE_NAME ::= STRING_LITERAL
This pragma indicates that the given entity is not used outside the compilation unit it is defined in. The entity may be either a subprogram or a variable.
If the entity to be eliminated is a library level subprogram, then
the first form of pragma Eliminate is used with only a single argument.
In this form, the Unit_Name argument specifies the name of the
library level unit to be eliminated.
In all other cases, both Unit_Name and Entity arguments
are required. item is an entity of a library package, then the first
argument specifies the unit name, and the second argument specifies
the particular entity. If the second argument is in string form, it must
correspond to the internal manner in which GNAT stores entity names (see
compilation unit Namet in the compiler sources for details).
The remaining parameters are optionally used to distinguish between overloaded subprograms. There are two ways of doing this.
Use Parameter_Types and Result_Type to specify the
profile of the subprogram to be eliminated in a manner similar to that
used for
the extended Import and Export pragmas, except that the
subtype names are always given as string literals, again corresponding
to the internal manner in which GNAT stores entity names.
Alternatively, the Homonym_Number parameter is used to specify
which overloaded alternative is to be eliminated. A value of 1 indicates
the first subprogram (in lexical order), 2 indicates the second etc.
The effect of the pragma is to allow the compiler to eliminate the code or data associated with the named entity. Any reference to an eliminated entity outside the compilation unit it is defined in, causes a compile time or link time error.
The parameters of this pragma may be given in any order, as long as the usual rules for use of named parameters and position parameters are used.
The intention of pragma Eliminate is to allow a program to be compiled
in a system independent manner, with unused entities eliminated, without
the requirement of modifying the source text. Normally the required set
of Eliminate pragmas is constructed automatically using the gnatelim tool.
Elimination of unused entities local to a compilation unit is automatic,
without requiring the use of pragma Eliminate.
Note that the reason this pragma takes string literals where names might
be expected is that a pragma Eliminate can appear in a context where the
relevant names are not visible.
pragma Export_Exception pragma Export_Exception (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL,]
[, [Form =>] Ada | VMS]
[, [Code =>] static_integer_EXPRESSION]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
This pragma is implemented only in the OpenVMS implementation of GNAT. It causes the specified exception to be propagated outside of the Ada program, so that it can be handled by programs written in other OpenVMS languages. This pragma establishes an external name for an Ada exception and makes the name available to the OpenVMS Linker as a global symbol. For further details on this pragma, see the DEC Ada Language Reference Manual, section 13.9a3.2.
pragma Export_Function ... pragma Export_Function (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Parameter_Types =>] PARAMETER_TYPES]
[, [Result_Type =>] result_SUBTYPE_MARK]
[, [Mechanism =>] MECHANISM]
[, [Result_Mechanism =>] MECHANISM_NAME]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
PARAMETER_TYPES ::=
null
| SUBTYPE_MARK {, SUBTYPE_MARK}
MECHANISM ::=
MECHANISM_NAME
| (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
MECHANISM_ASSOCIATION ::=
[formal_parameter_NAME =>] MECHANISM_NAME
MECHANISM_NAME ::=
Value
| Reference
| Descriptor [([Class =>] CLASS_NAME)]
CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca
Use this pragma to make a function externally callable and optionally
provide information on mechanisms to be used for passing parameter and
result values. We recommend, for the purposes of improving portability,
this pragma always be used in conjunction with a separate pragma
Export, which must precede the pragma Export_Function.
GNAT does not require a separate pragma Export, but if none is
present, Convention Ada is assumed, which is usually
not what is wanted, so it is usually appropriate to use this
pragma in conjunction with a Export or Convention
pragma that specifies the desired foreign convention.
Pragma Export_Function
(and Export, if present) must appear in the same declarative
region as the function to which they apply.
internal_name must uniquely designate the function to which the
pragma applies. If more than one function name exists of this name in
the declarative part you must use the Parameter_Types and
Result_Type parameters is mandatory to achieve the required
unique designation. subtype_ marks in these parameters must
exactly match the subtypes in the corresponding function specification,
using positional notation to match parameters with subtype marks.
Passing by descriptor is supported only on the OpenVMS ports of GNAT.
pragma Export_Object ... pragma Export_Object
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Size =>] EXTERNAL_SYMBOL]
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
This pragma designates an object as exported, and apart from the
extended rules for external symbols, is identical in effect to the use of
the normal Export pragma applied to an object. You may use a
separate Export pragma (and you probably should from the point of view
of portability), but it is not required. Size is syntax checked,
but otherwise ignored by GNAT.
pragma Export_Procedure ... pragma Export_Procedure (
[Internal =>] LOCAL_NAME
[, [External =>] EXTERNAL_SYMBOL]
[, [Parameter_Types =>] PARAMETER_TYPES]
[, [Mechanism =>] MECHANISM]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
PARAMETER_TYPES ::=
null
| SUBTYPE_MARK {, SUBTYPE_MARK}
MECHANISM ::=
MECHANISM_NAME
| (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
MECHANISM_ASSOCIATION ::=
[formal_parameter_NAME =>] MECHANISM_NAME
MECHANISM_NAME ::=
Value
| Reference
| Descriptor [([Class =>] CLASS_NAME)]
CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca
This pragma is identical to Export_Function except that it
applies to a procedure rather than a function and the parameters
Result_Type and Result_Mechanism are not permitted.
GNAT does not require a separate pragma Export, but if none is
present, Convention Ada is assumed, which is usually
not what is wanted, so it is usually appropriate to use this
pragma in conjunction with a Export or Convention
pragma that specifies the desired foreign convention.
pragma Export_Valued_Procedure pragma Export_Valued_Procedure (
[Internal =>] LOCAL_NAME
[, [External =>] EXTERNAL_SYMBOL]
[, [Parameter_Types =>] PARAMETER_TYPES]
[, [Mechanism =>] MECHANISM]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
PARAMETER_TYPES ::=
null
| SUBTYPE_MARK {, SUBTYPE_MARK}
MECHANISM ::=
MECHANISM_NAME
| (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
MECHANISM_ASSOCIATION ::=
[formal_parameter_NAME =>] MECHANISM_NAME
MECHANISM_NAME ::=
Value
| Reference
| Descriptor [([Class =>] CLASS_NAME)]
CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca
This pragma is identical to Export_Procedure except that the
first parameter of local_name, which must be present, must be of
mode OUT, and externally the subprogram is treated as a function
with this parameter as the result of the function. GNAT provides for
this capability to allow the use of OUT and IN OUT
parameters in interfacing to external functions (which are not permitted
in Ada functions).
GNAT does not require a separate pragma Export, but if none is
present, Convention Ada is assumed, which is almost certainly
not what is wanted since the whole point of this pragma is to interface
with foreign language functions, so it is usually appropriate to use this
pragma in conjunction with a Export or Convention
pragma that specifies the desired foreign convention.
pragma Extend_System pragma Extend_System ([Name =>] IDENTIFIER);
This pragma is used to provide backwards compatibility with other
implementations that extend the facilities of package System. In
GNAT, System contains only the definitions that are present in
the Ada 95 RM. However, other implementations, notably the DEC Ada 83
implementation, provide many extensions to package System.
For each such implementation accommodated by this pragma, GNAT provides a
package Aux_xxx, e.g. Aux_DEC for the DEC Ada 83
implementation, which provides the required additional definitions. You
can use this package in two ways. You can with it in the normal
way and access entities either by selection or using a use
clause. In this case no special processing is required.
However, if existing code contains references such as
System.xxx where xxx is an entity in the extended
definitions provided in package System, you may use this pragma
to extend visibility in System in a non-standard way that
provides greater compatibility with the existing code. Pragma
Extend_System is a configuration pragma whose single argument is
the name of the package containing the extended definition
(e.g. Aux_DEC for the DEC Ada case). A unit compiled under
control of this pragma will be processed using special visibility
processing that looks in package System.Aux_xxx where
Aux_xxx is the pragma argument for any entity referenced in
package System, but not found in package System.
You can use this pragma either to access a predefined System
extension supplied with the compiler, for example Aux_DEC or
you can construct your own extension unit following the above
definition. Note that such a package is a child of System
and thus is considered part of the implementation. To compile
it you will have to use the appropriate switch for compiling
system units. See the GNAT User's Guide for details.
pragma External pragma External (
[ Convention =>] convention_IDENTIFIER,
[ Entity =>] local_NAME
[, [External_Name =>] static_string_EXPRESSION ]
[, [Link_Name =>] static_string_EXPRESSION ]);
This pragma is identical in syntax and semantics to pragma
Export as defined in the Ada Reference Manual. It is
provided for compatibility with some Ada 83 compilers that
used this pragma for exactly the same purposes as pragma
Export before the latter was standardized.
pragma External_Name_Casing pragma External_Name_Casing (
Uppercase | Lowercase
[, Uppercase | Lowercase | As_Is]);
This pragma provides control over the casing of external names associated with Import and Export pragmas. There are two cases to consider:
pragma Import (C, C_Routine);
Since Ada is a case insensitive language, the spelling of the identifier in
the Ada source program does not provide any information on the desired
casing of the external name, and so a convention is needed. In GNAT the
default treatment is that such names are converted to all lower case
letters. This corresponds to the normal C style in many environments.
The first argument of pragma External_Name_Casing can be used to
control this treatment. If Uppercase is specified, then the name
will be forced to all uppercase letters. If Lowercase is specified,
then the normal default of all lower case letters will be used.
This same implicit treatment is also used in the case of extended DEC Ada 83
compatible Import and Export pragmas where an external name is explicitly
specified using an identifier rather than a string.
pragma Import (C, C_Routine, "C_routine");
In this case, the string literal normally provides the exact casing required
for the external name. The second argument of pragma
External_Name_Casing may be used to modify this behavior.
If Uppercase is specified, then the name
will be forced to all uppercase letters. If Lowercase is specified,
then the name will be forced to all lowercase letters. A specification of
As_Is provides the normal default behavior in which the casing is
taken from the string provided.
This pragma may appear anywhere that a pragma is valid. In particular, it can be used as a configuration pragma in the gnat.adc file, in which case it applies to all subsequent compilations, or it can be used as a program unit pragma, in which case it only applies to the current unit, or it can be used more locally to control individual Import/Export pragmas.
It is primarily intended for use with OpenVMS systems, where many compilers convert all symbols to upper case by default. For interfacing to such compilers (e.g. the DEC C compiler), it may be convenient to use the pragma:
pragma External_Name_Casing (Uppercase, Uppercase);
to enforce the upper casing of all external symbols.
pragma Finalize_Storage_Only pragma Finalize_Storage_Only (first_subtype_LOCAL_NAME);
This pragma allows the compiler not to emit a Finalize call for objects defined at the library level. This is mostly useful for types where finalization is only used to deal with storage reclamation since in most environments it is not necessary to reclaim memory just before terminating execution, hence the name.
pragma Float_Representation pragma Float_Representation (FLOAT_REP);
FLOAT_REP ::= VAX_Float | IEEE_Float
This pragma is implemented only in the OpenVMS implementation of GNAT.
It allows control over the internal representation chosen for the predefined
floating point types declared in the packages Standard and
System. For further details on this pragma, see the
DEC Ada Language Reference Manual, section 3.5.7a. Note that to use this
pragma, the standard runtime libraries must be recompiled. See the
description of the GNAT LIBRARY command in the OpenVMS version
of the GNAT Users Guide for details on the use of this command.
pragma Ident pragma Ident (static_string_EXPRESSION);
This pragma provides a string identification in the generated object file, if the system supports the concept of this kind of identification string. The maximum permitted length of the string literal is 31 characters. This pragma is allowed only in the outermost declarative part or declarative items of a compilation unit. On OpenVMS systems, the effect of the pragma is identical to the effect of the DEC Ada 83 pragma of the same name.
pragma Import_Exception pragma Import_Exception (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL,]
[, [Form =>] Ada | VMS]
[, [Code =>] static_integer_EXPRESSION]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
This pragma is implemented only in the OpenVMS implementation of GNAT. It allows OpenVMS conditions (for example, from OpenVMS system services or other OpenVMS languages) to be propagated to Ada programs as Ada exceptions. The pragma specifies that the exception associated with an exception declaration in an Ada program be defined externally (in non-Ada code). For further details on this pragma, see the DEC Ada Language Reference Manual, section 13.9a.3.1.
pragma Import_Function ... pragma Import_Function (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Parameter_Types =>] PARAMETER_TYPES]
[, [Result_Type =>] SUBTYPE_MARK]
[, [Mechanism =>] MECHANISM]
[, [Result_Mechanism =>] MECHANISM_NAME]
[, [First_Optional_Parameter =>] IDENTIFIER]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
PARAMETER_TYPES ::=
null
| SUBTYPE_MARK {, SUBTYPE_MARK}
MECHANISM ::=
MECHANISM_NAME
| (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
MECHANISM_ASSOCIATION ::=
[formal_parameter_NAME =>] MECHANISM_NAME
MECHANISM_NAME ::=
Value
| Reference
| Descriptor [([Class =>] CLASS_NAME)]
CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca
This pragma is used in conjunction with a pragma Import to
specify additional information for an imported function. The pragma
Import (or equivalent pragma Interface) must precede the
Import_Function pragma and both must appear in the same
declarative part as the function specification.
The Internal_Name argument must uniquely designate
the function to which the
pragma applies. If more than one function name exists of this name in
the declarative part you must use the Parameter_Types and
Result_Type parameters to achieve the required unique
designation. Subtype marks in these parameters must exactly match the
subtypes in the corresponding function specification, using positional
notation to match parameters with subtype marks.
You may optionally use the Mechanism and Result_Mechanism parameters to specify passing mechanisms for the parameters and result. If you specify a single mechanism name, it applies to all parameters. Otherwise you may specify a mechanism on a parameter by parameter basis using either positional or named notation. If the mechanism is not specified, the default mechanism is used.
Passing by descriptor is supported only on the to OpenVMS ports of GNAT.
First_Optional_Parameter applies only to OpenVMS ports of GNAT.
It specifies that the designated parameter and all following parameters
are optional, meaning that they are not passed at the generated code
level (this is distinct from the notion of optional parameters in Ada
where the parameters are passed anyway with the designated optional
parameters). All optional parameters must be of mode IN and have
default parameter values that are either known at compile time
expressions, or uses of the 'Null_Parameter attribute.
pragma Import_Object pragma Import_Object
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL],
[, [Size =>] EXTERNAL_SYMBOL])
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
This pragma designates an object as imported, and apart from the
extended rules for external symbols, is identical in effect to the use of
the normal Import pragma applied to an object. Unlike the
subprogram case, you need not use a separate Import pragma,
although you may do so (and probably should do so from a portability
point of view). size is syntax checked, but otherwise ignored by
GNAT.
pragma Import_Procedure pragma Import_Procedure (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Parameter_Types =>] PARAMETER_TYPES]
[, [Mechanism =>] MECHANISM]
[, [First_Optional_Parameter =>] IDENTIFIER]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
PARAMETER_TYPES ::=
null
| SUBTYPE_MARK {, SUBTYPE_MARK}
MECHANISM ::=
MECHANISM_NAME
| (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
MECHANISM_ASSOCIATION ::=
[formal_parameter_NAME =>] MECHANISM_NAME
MECHANISM_NAME ::=
Value
| Reference
| Descriptor [([Class =>] CLASS_NAME)]
CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca
This pragma is identical to Import_Function except that it
applies to a procedure rather than a function and the parameters
Result_Type and Result_Mechanism are not permitted.
pragma Import_Valued_Procedure ... pragma Import_Valued_Procedure (
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Parameter_Types =>] PARAMETER_TYPES]
[, [Mechanism =>] MECHANISM]
[, [First_Optional_Parameter =>] IDENTIFIER]);
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
PARAMETER_TYPES ::=
null
| SUBTYPE_MARK {, SUBTYPE_MARK}
MECHANISM ::=
MECHANISM_NAME
| (MECHANISM_ASSOCIATION {, MECHANISM_ASSOCIATION})
MECHANISM_ASSOCIATION ::=
[formal_parameter_NAME =>] MECHANISM_NAME
MECHANISM_NAME ::=
Value
| Reference
| Descriptor [([Class =>] CLASS_NAME)]
CLASS_NAME ::= ubs | ubsb | uba | s | sb | a | nca
This pragma is identical to Import_Procedure except that the
first parameter of local_name, which must be present, must be of
mode OUT, and externally the subprogram is treated as a function
with this parameter as the result of the function. The purpose of this
capability is to allow the use of OUT and IN OUT
parameters in interfacing to external functions (which are not permitted
in Ada functions). You may optionally use the Mechanism
parameters to specify passing mechanisms for the parameters.
If you specify a single mechanism name, it applies to all parameters.
Otherwise you may specify a mechanism on a parameter by parameter
basis using either positional or named notation. If the mechanism is not
specified, the default mechanism is used.
Note that it is important to use this pragma in conjunction with a separate pragma Import that specifies the desired convention, since otherwise the default convention is Ada, which is almost certainly not what is required.
pragma Initialize_Scalars pragma Initialize_Scalars;
This pragma is similar to Normalize_Scalars conceptually but has
two important differences. First, there is no requirement for the pragma
to be used uniformly in all units of a partition, in particular, it is fine
to use this just for some or all of the application units of a partition,
without needing to recompile the run-time library.
In the case where some units are compiled with the pragma, and some without, then a declaration of a variable where the type is defined in package Standard or is locally declared will always be subject to initialization, as will any declaration of a scalar variable. For composite variables, whether the variable is initialized may also depend on whether the package in which the type of the variable is declared is compiled with the pragma.
The other important difference is that there is control over the value used for initializing scalar objects. At bind time, you can select whether to initialize with invalid values (like Normalize_Scalars), or with high or low values, or with a specified bit pattern. See the users guide for binder options for specifying these cases.
This means that you can compile a program, and then without having to recompile the program, you can run it with different values being used for initializing otherwise uninitialized values, to test if your program behavior depends on the choice. Of course the behavior should not change, and if it does, then most likely you have an erroneous reference to an uninitialized value.
Note that pragma Initialize_Scalars is particularly useful in
conjunction with the enhanced validity checking that is now provided
in GNAT, which checks for invalid values under more conditions.
Using this feature (see description of the -gnatv flag in the
users guide) in conjunction with pragma Initialize_Scalars
provides a powerful new tool to assist in the detection of problems
caused by uninitialized variables.
pragma Inline_Always pragma Inline_Always (NAME [, NAME]);
Similar to pragma Inline except that inlining is not subject to
the use of option -gnatn for inter-unit inlining.
pragma Inline_Generic pragma Inline_Generic (generic_package_NAME)
This is implemented for compatibility with DEC Ada 83 and is recognized, but otherwise ignored, by GNAT. All generic instantiations are inlined by default when using GNAT.
pragma Interface pragma Interface (
[Convention =>] convention_identifier,
[Entity =>] local_name
[, [External_Name =>] static_string_expression],
[, [Link_Name =>] static_string_expression]);
This pragma is identical in syntax and semantics to
the standard Ada 95 pragma Import. It is provided for compatibility
with Ada 83. The definition is upwards compatible both with pragma
Interface as defined in the Ada 83 Reference Manual, and also
with some extended implementations of this pragma in certain Ada 83
implementations.
pragma Interface_Name pragma Interface_Name (
[Entity =>] LOCAL_NAME
[, [External_Name =>] static_string_EXPRESSION]
[, [Link_Name =>] static_string_EXPRESSION]);
This pragma provides an alternative way of specifying the interface name for an interfaced subprogram, and is provided for compatibility with Ada 83 compilers that use the pragma for this purpose. You must provide at least one of External_Name or Link_Name.
pragma License pragma License (Unrestricted | GPL | Modified_GPL | Restricted);
This pragma is provided to allow automated checking for appropriate license
conditions with respect to the standard and modified GPL. A pragma License,
which is a configuration pragma that typically appears at the start of a
source file or in a separate gnat.adc file, specifies the licensing
conditions of a unit as follows:
with'ed by a restricted unit.
with units
which are licensed under the modified GPL (this is the whole point of the
modified GPL).
Normally a unit with no License pragma is considered to have an
unknown license, and no checking is done. However, standard GNAT headers
are recognized, and license information is derived from them as follows.
If the string “GNU General Public License” is found, then the unit is assumed to have GPL license, unless the string “As a special exception” follows, in which case the license is assumed to be modified GPL.
If one of the strings “This specification is adapated from the Ada Semantic Interface” or “This specification is derived from the Ada Reference Manual” is found then the unit is assumed to be unrestricted.
These default actions means that a program with a restricted license pragma
will automatically get warnings if a GPL unit is inappropriately
with'ed. For example, the program:
with Sem_Ch3;
with GNAT.Sockets;
procedure Secret_Stuff is
...
end Secret_Stuff
if compiled with pragma License (Restricted) in a
gnat.adc file will generate the warning:
1. with Sem_Ch3;
|
>>> license of withed unit "Sem_Ch3" is incompatible
2. with GNAT.Sockets;
3. procedure Secret_Stuff is
Here we get a warning on Sem_Ch3 since it is part of the GNAT
compiler and is licensed under the
GPL, but no warning for GNAT.Sockets which is part of the GNAT
run time, and is therefore licensed under the modified GPL.
pragma Link_With pragma Link_With (static_string_EXPRESSION {,static_string_EXPRESSION});
This pragma is provided for compatibility with certain Ada 83 compilers.
It has exactly the same effect as pragma Linker_Options except
that spaces occurring within one of the string expressions are treated
as separators. For example, in the following case:
pragma Link_With ("-labc -ldef");
results in passing the strings -labc and -ldef as two
separate arguments to the linker. In addition pragma Link_With allows
multiple arguments, with the same effect as successive pragmas.
pragma Linker_Alias pragma Linker_Alias (
[Entity =>] LOCAL_NAME
[Alias =>] static_string_EXPRESSION);
This pragma establishes a linker alias for the given named entity. For further details on the exact effect, consult the GCC manual.
pragma Linker_Section pragma Linker_Section (
[Entity =>] LOCAL_NAME
[Section =>] static_string_EXPRESSION);
This pragma specifies the name of the linker section for the given entity. For further details on the exact effect, consult the GCC manual.
pragma No_Run_Time pragma No_Run_Time;
This is a configuration pragma that makes sure the user code does not use nor need anything from the GNAT run time. This is mostly useful in context where code certification is required. Please consult the GNAT Pro High-Integrity Edition User's Guide for additional information.
pragma Normalize_Scalars pragma Normalize_Scalars;
This is a language defined pragma which is fully implemented in GNAT. The effect is to cause all scalar objects that are not otherwise initialized to be initialized. The initial values are implementation dependent and are as follows:
Standard.CharacterStandard.Wide_CharacterInteger types subtype Ityp is integer range 1 .. 10;
then objects of type x will be initialized to Integer'First, a negative
number that is certainly outside the range of subtype Ityp.
Real typesModular typesEnumeration types2 ** typ'Size - 1. This will be out of range of the enumeration
subtype in all cases except where the subtype contains exactly
2**8, 2**16, or 2**32 elements.
pragma Long_Float pragma Long_Float (FLOAT_FORMAT);
FLOAT_FORMAT ::= D_Float | G_Float
This pragma is implemented only in the OpenVMS implementation of GNAT.
It allows control over the internal representation chosen for the predefined
type Long_Float and for floating point type representations with
digits specified in the range 7 through 15.
For further details on this pragma, see the
DEC Ada Language Reference Manual, section 3.5.7b. Note that to use this
pragma, the standard runtime libraries must be recompiled. See the
description of the GNAT LIBRARY command in the OpenVMS version
of the GNAT User's Guide for details on the use of this command.
pragma Machine_Attribute ... pragma Machine_Attribute (
[Attribute_Name =>] string_EXPRESSION,
[Entity =>] LOCAL_NAME);
Machine dependent attributes can be specified for types and/or
declarations. Currently only subprogram entities are supported. This
pragma is semantically equivalent to
__attribute__((string_expression)) in GNU C,
where string_expression is
recognized by the GNU C macros VALID_MACHINE_TYPE_ATTRIBUTE and
VALID_MACHINE_DECL_ATTRIBUTE which are defined in the
configuration header file tm.h for each machine. See the GCC
manual for further information.
pragma Main_Storage pragma Main_Storage
(MAIN_STORAGE_OPTION [, MAIN_STORAGE_OPTION]);
MAIN_STORAGE_OPTION ::=
[WORKING_STORAGE =>] static_SIMPLE_EXPRESSION
| [TOP_GUARD =>] static_SIMPLE_EXPRESSION
This pragma is provided for compatibility with OpenVMS Vax Systems. It has no effect in GNAT, other than being syntax checked. Note that the pragma also has no effect in DEC Ada 83 for OpenVMS Alpha Systems.
pragma No_Return pragma No_Return (procedure_LOCAL_NAME);
procedure_local_NAME must refer to one or more procedure
declarations in the current declarative part. A procedure to which this
pragma is applied may not contain any explicit return statements,
and also may not contain any implicit return statements from falling off
the end of a statement sequence. One use of this pragma is to identify
procedures whose only purpose is to raise an exception.
Another use of this pragma is to suppress incorrect warnings about missing returns in functions, where the last statement of a function statement sequence is a call to such a procedure.
pragma Passive pragma Passive ([Semaphore | No]);
Syntax checked, but otherwise ignored by GNAT. This is recognized for
compatibility with DEC Ada 83 implementations, where it is used within a
task definition to request that a task be made passive. If the argument
Semaphore is present, or no argument is omitted, then DEC Ada 83
treats the pragma as an assertion that the containing task is passive
and that optimization of context switch with this task is permitted and
desired. If the argument No is present, the task must not be
optimized. GNAT does not attempt to optimize any tasks in this manner
(since protected objects are available in place of passive tasks).
pragma Polling pragma Polling (ON | OFF);
This pragma controls the generation of polling code. This is normally off.
If pragma Polling (ON) is used then periodic calls are generated to
the routine Ada.Exceptions.Poll. This routine is a separate unit in the
runtime library, and can be found in file a-excpol.adb.
Pragma Polling can appear as a configuration pragma (for example it can be
placed in the gnat.adc file) to enable polling globally, or it can be used
in the statement or declaration sequence to control polling more locally.
A call to the polling routine is generated at the start of every loop and
at the start of every subprogram call. This guarantees that the Poll
routine is called frequently, and places an upper bound (determined by
the complexity of the code) on the period between two Poll calls.
The primary purpose of the polling interface is to enable asynchronous
aborts on targets that cannot otherwise support it (for example Windows
NT), but it may be used for any other purpose requiring periodic polling.
The standard version is null, and can be replaced by a user program. This
will require re-compilation of the Ada.Exceptions package that can be found
in files a-except.ads and a-except.adb.
A standard alternative unit (in file 4wexcpol.adb in the standard GNAT
distribution) is used to enable the asynchronous abort capability on
targets that do not normally support the capability. The version of Poll
in this file makes a call to the appropriate runtime routine to test for
an abort condition.
Note that polling can also be enabled by use of the -gnatP switch. See
the GNAT User's Guide for details.
pragma Propagate_Exceptions pragma Propagate_Exceptions (subprogram_LOCAL_NAME);
This pragma indicates that the given entity, which is the name of an
imported foreign-language subprogram may receive an Ada exception,
and that the exception should be propagated. It is relevant only if
zero cost exception handling is in use, and is thus never needed if
the alternative longjmp / setjmp implementation of exceptions is used
(although it is harmless to use it in such cases).
The implementation of fast exceptions always properly propagates
exceptions through Ada code, as described in the Ada Reference Manual.
However, this manual is silent about the propagation of exceptions
through foreign code. For example, consider the
situation where P1 calls
P2, and P2 calls P3, where
P1 and P3 are in Ada, but P2 is in C.
P3 raises an Ada exception. The question is whether or not
it will be propagated through P2 and can be handled in
P1.
For the longjmp / setjmp implementation of exceptions, the answer is
always yes. For some targets on which zero cost exception handling
is implemented, the answer is also always yes. However, there are
some targets, notably in the current version all x86 architecture
targets, in which the answer is that such propagation does not
happen automatically. If such propagation is required on these
targets, it is mandatory to use Propagate_Exceptions to
name all foreign language routines through which Ada exceptions
may be propagated.
pragma Psect_Object pragma Psect_Object
[Internal =>] LOCAL_NAME,
[, [External =>] EXTERNAL_SYMBOL]
[, [Size =>] EXTERNAL_SYMBOL]
EXTERNAL_SYMBOL ::=
IDENTIFIER
| static_string_EXPRESSION
This pragma is identical in effect to pragma Common_Object.
pragma Pure_Function pragma Pure_Function ([Entity =>] function_LOCAL_NAME);
This pragma appears in the same declarative part as a function
declaration (or a set of function declarations if more than one
overloaded declaration exists, in which case the pragma applies
to all entities). If specifies that the function Entity is
to be considered pure for the purposes of code generation. This means
that the compiler can assume that there are no side effects, and
in particular that two calls with identical arguments produce the
same result. It also means that the function can be used in an
address clause.
Note that, quite deliberately, there are no static checks to try
to ensure that this promise is met, so Pure_Function can be used
with functions that are conceptually pure, even if they do modify
global variables. For example, a square root function that is
instrumented to count the number of times it is called is still
conceptually pure, and can still be optimized, even though it
modifies a global variable (the count). Memo functions are another
example (where a table of previous calls is kept and consulted to
avoid re-computation).
Note: Most functions in a Pure package are automatically pure, and
there is no need to use pragma Pure_Function for such functions. An
exception is any function that has at least one formal of type
System.Address or a type derived from it. Such functions are not
considered pure by default, since the compiler assumes that the
Address parameter may be functioning as a pointer and that the
referenced data may change even if the address value does not. The use
of pragma Pure_Function for such a function will override this default
assumption, and cause the compiler to treat such a function as pure.
Note: If pragma Pure_Function is applied to a renamed function, it
applies to the underlying renamed function. This can be used to
disambiguate cases of overloading where some but not all functions
in a set of overloaded functions are to be designated as pure.
pragma Ravenscar pragma Ravenscar
A configuration pragma that establishes the following set of restrictions:
No_Abort_StatementsNo_Select_StatementsNo_Task_HierarchyNo_Task_AllocatorsNo_Dynamic_PrioritiesNo_Terminate_AlternativesNo_Dynamic_InterruptsNo_Protected_Type_AllocatorsNo_Local_Protected_ObjectsNo_RequeueNo_CalendarNo_Relative_DelayNo_Task_AttributesStatic_Storage_SizeBoolean_Entry_BarriersMax_Asynchronous_Select_Nesting = 0Max_Task_Entries = 0Max_Protected_Entries = 1Max_Select_Alternatives = 0No_Task_TerminationNo_Entry_QueueThis set of restrictions corresponds to the definition of the “Ravenscar Profile” for limited tasking, devised and published by the International Real-Time Ada Workshop, 1997.
The above set is a superset of the restrictions provided by pragma
Restricted_Run_Time, it includes six additional restrictions
(Boolean_Entry_Barriers, No_Select_Statements,
No_Calendar, Static_Storage_Size,
No_Relative_Delay and No_Task_Termination). This means
that pragma Ravenscar, like the pragma Restricted_Run_Time, automatically
causes the use of a simplified, more efficient version of the tasking
run-time system.
pragma Restricted_Run_Time pragma Restricted_Run_Time
A configuration pragma that establishes the following set of restrictions:
This set of restrictions causes the automatic selection of a simplified version of the run time that provides improved performance for the limited set of tasking functionality permitted by this set of restrictions.
pragma Share_Generic pragma Share_Generic (NAME {, NAME});
This pragma is recognized for compatibility with other Ada compilers but is ignored by GNAT. GNAT does not provide the capability for sharing of generic code. All generic instantiations result in making an inlined copy of the template with appropriate substitutions.
pragma Source_File_Name pragma Source_File_Name (
[Unit_Name =>] unit_NAME,
Spec_File_Name => STRING_LITERAL);
pragma Source_File_Name (
[Unit_Name =>] unit_NAME,
Body_File_Name => STRING_LITERAL);
Use this to override the normal naming convention. It is a configuration pragma, and so has the usual applicability of configuration pragmas (i.e. it applies to either an entire partition, or to all units in a compilation, or to a single unit, depending on how it is used. unit_name is mapped to file_name_literal. The identifier for the second argument is required, and indicates whether this is the file name for the spec or for the body.
Another form of the Source_File_Name pragma allows
the specification of patterns defining alternative file naming schemes
to apply to all files.
pragma Source_File_Name
(Spec_File_Name => STRING_LITERAL
[,Casing => CASING_SPEC]
[,Dot_Replacement => STRING_LITERAL]);
pragma Source_File_Name
(Body_File_Name => STRING_LITERAL
[,Casing => CASING_SPEC]
[,Dot_Replacement => STRING_LITERAL]);
pragma Source_File_Name
(Subunit_File_Name => STRING_LITERAL
[,Casing => CASING_SPEC]
[,Dot_Replacement => STRING_LITERAL]);
CASING_SPEC ::= Lowercase | Uppercase | Mixedcase
The first argument is a pattern that contains a single asterisk indicating the point at which the unit name is to be inserted in the pattern string to form the file name. The second argument is optional. If present it specifies the casing of the unit name in the resulting file name string. The default is lower case. Finally the third argument allows for systematic replacement of any dots in the unit name by the specified string literal.
For more details on the use of the Source_File_Name pragma,
see the sections “Using Other File Names” and
“Alternative File Naming Schemes” in the GNAT User's Guide.
pragma Source_Reference pragma Source_Reference (INTEGER_LITERAL,
STRING_LITERAL);
This pragma must appear as the first line of a source file.
integer_literal is the logical line number of the line following
the pragma line (for use in error messages and debugging
information). string_literal is a static string constant that
specifies the file name to be used in error messages and debugging
information. This is most notably used for the output of gnatchop
with the -r switch, to make sure that the original unchopped
source file is the one referred to.
The second argument must be a string literal, it cannot be a static string expression other than a string literal. This is because its value is needed for error messages issued by all phases of the compiler.
pragma Stream_Convert pragma Stream_Convert (
[Entity =>] type_LOCAL_NAME,
[Read =>] function_NAME,
[Write =>] function NAME);
This pragma provides an efficient way of providing stream functions for types defined in packages. Not only is it simpler to use than declaring the necessary functions with attribute representation clauses, but more significantly, it allows the declaration to made in such a way that the stream packages are not loaded unless they are needed. The use of the Stream_Convert pragma adds no overhead at all, unless the stream attributes are actually used on the designated type.
The first argument specifies the type for which stream functions are provided. The second parameter provides a function used to read values of this type. It must name a function whose argument type may be any subtype, and whose returned type must be the type given as the first argument to the pragma.
The meaning of the Read parameter is that if a stream attribute directly or indirectly specifies reading of the type given as the first parameter, then a value of the type given as the argument to the Read function is read from the stream, and then the Read function is used to convert this to the required target type.
Similarly the Write parameter specifies how to treat write attributes that directly or indirectly apply to the type given as the first parameter. It must have an input parameter of the type specified by the first parameter, and the return type must be the same as the input type of the Read function. The effect is to first call the Write function to convert to the given stream type, and then write the result type to the stream.
The Read and Write functions must not be overloaded subprograms. If necessary renamings can be supplied to meet this requirement. The usage of this attribute is best illustrated by a simple example, taken from the GNAT implementation of package Ada.Strings.Unbounded:
function To_Unbounded (S : String)
return Unbounded_String
renames To_Unbounded_String;
pragma Stream_Convert
(Unbounded_String, To_Unbounded, To_String);
The specifications of the referenced functions, as given in the Ada 95 Reference Manual are:
function To_Unbounded_String (Source : String)
return Unbounded_String;
function To_String (Source : Unbounded_String)
return String;
The effect is that if the value of an unbounded string is written to a
stream, then the representation of the item in the stream is in the same
format used for Standard.String, and this same representation is
expected when a value of this type is read from the stream.
pragma Style_Checks pragma Style_Checks (string_LITERAL | ALL_CHECKS |
On | Off [, LOCAL_NAME]);
This pragma is used in conjunction with compiler switches to control the built in style checking provided by GNAT. The compiler switches, if set provide an initial setting for the switches, and this pragma may be used to modify these settings, or the settings may be provided entirely by the use of the pragma. This pragma can be used anywhere that a pragma is legal, including use as a configuration pragma (including use in the gnat.adc file).
The form with a string literal specifies which style options are to be
activated. These are additive, so they apply in addition to any previously
set style check options. The codes for the options are the same as those
used in the -gnaty switch to gcc or gnatmake.
For example the following two methods can be used to enable
layout checking:
pragma Style_Checks ("l");
gcc -c -gnatyl ...
The form ALL_CHECKS activates all standard checks (its use is equivalent
to the use of the gnaty switch with no options. See GNAT User's
Guide for details.
The forms with Off and On
can be used to temporarily disable style checks
as shown in the following example:
pragma Style_Checks ("k"); -- requires keywords in lower case
pragma Style_Checks (Off); -- turn off style checks
NULL; -- this will not generate an error message
pragma Style_Checks (On); -- turn style checks back on
NULL; -- this will generate an error message
Finally the two argument form is allowed only if the first argument is
On or Off. The effect is to turn of semantic style checks
for the specified entity, as shown in the following example:
pragma Style_Checks ("r"); -- require consistency of identifier casing
Arg : Integer;
Rf1 : Integer := ARG; -- incorrect, wrong case
pragma Style_Checks (Off, Arg);
Rf2 : Integer := ARG; -- OK, no error
pragma Subtitle pragma Subtitle ([Subtitle =>] STRING_LITERAL);
This pragma is recognized for compatibility with other Ada compilers but is ignored by GNAT.
pragma Suppress_All pragma Suppress_All;
This pragma can only appear immediately following a compilation
unit. The effect is to apply Suppress (All_Checks) to the unit
which it follows. This pragma is implemented for compatibility with DEC
Ada 83 usage. The use of pragma Suppress (All_Checks) as a normal
configuration pragma is the preferred usage in GNAT.
pragma Suppress_Initialization pragma Suppress_Initialization ([Entity =>] type_Name);
This pragma suppresses any implicit or explicit initialization associated with the given type name for all variables of this type.
pragma Task_Info pragma Task_Info (EXPRESSION);
This pragma appears within a task definition (like pragma
Priority) and applies to the task in which it appears. The
argument must be of type System.Task_Info.Task_Info_Type.
The Task_Info pragma provides system dependent control over
aspect of tasking implementation, for example, the ability to map
tasks to specific processors. For details on the facilities available
for the version of GNAT that you are using, see the documentation
in the specification of package System.Task_Info in the runtime
library.
pragma Task_Name pragma Task_Name (string_EXPRESSION);
This pragma appears within a task definition (like pragma
Priority) and applies to the task in which it appears. The
argument must be of type String, and provides a name to be used for
the task instance when the task is created. Note that this expression
is not required to be static, and in particular, it can contain
references to task discriminants. This facility can be used to
provide different names for different tasks as they are created,
as illustrated in the example below.
The task name is recorded internally in the run-time structures
and is accessible to tools like the debugger. In addition the
routine Ada.Task_Identification.Image will return this
string, with a unique task address appended.
-- Example of the use of pragma Task_Name
with Ada.Task_Identification;
use Ada.Task_Identification;
with Text_IO; use Text_IO;
procedure t3 is
type Astring is access String;
task type Task_Typ (Name : access String) is
pragma Task_Name (Name.all);
end Task_Typ;
task body Task_Typ is
Nam : constant String := Image (Current_Task);
begin
Put_Line ("-->" & Nam (1 .. 14) & "<--");
end Task_Typ;
type Ptr_Task is access Task_Typ;
Task_Var : Ptr_Task;
begin
Task_Var :=
new Task_Typ (new String'("This is task 1"));
Task_Var :=
new Task_Typ (new String'("This is task 2"));
end;
pragma Task_Storage pragma Task_Storage
[Task_Type =>] LOCAL_NAME,
[Top_Guard =>] static_integer_EXPRESSION);
This pragma specifies the length of the guard area for tasks. The guard
area is an additional storage area allocated to a task. A value of zero
means that either no guard area is created or a minimal guard area is
created, depending on the target. This pragma can appear anywhere a
Storage_Size attribute definition clause is allowed for a task
type.
pragma Time_Slice pragma Time_Slice (static_duration_EXPRESSION);
For implementations of GNAT on operating systems where it is possible to supply a time slice value, this pragma may be used for this purpose. It is ignored if it is used in a system that does not allow this control, or if it appears in other than the main program unit. Note that the effect of this pragma is identical to the effect of the DEC Ada 83 pragma of the same name when operating under OpenVMS systems.
pragma Title pragma Title (TITLING_OPTION [, TITLING OPTION]);
TITLING_OPTION ::=
[Title =>] STRING_LITERAL,
| [Subtitle =>] STRING_LITERAL
Syntax checked but otherwise ignored by GNAT. This is a listing control pragma used in DEC Ada 83 implementations to provide a title and/or subtitle for the program listing. The program listing generated by GNAT does not have titles or subtitles.
Unlike other pragmas, the full flexibility of named notation is allowed for this pragma, i.e. the parameters may be given in any order if named notation is used, and named and positional notation can be mixed following the normal rules for procedure calls in Ada.
pragma Unchecked_Union pragma Unchecked_Union (first_subtype_LOCAL_NAME)
This pragma is used to declare that the specified type should be represented in a manner equivalent to a C union type, and is intended only for use in interfacing with C code that uses union types. In Ada terms, the named type must obey the following rules:
In addition, given a type that meets the above requirements, the following restrictions apply to its use throughout the program:
Equality and inequality operations on unchecked_unions are not
available, since there is no discriminant to compare and the compiler
does not even know how many bits to compare. It is implementation
dependent whether this is detected at compile time as an illegality or
whether it is undetected and considered to be an erroneous construct. In
GNAT, a direct comparison is illegal, but GNAT does not attempt to catch
the composite case (where two composites are compared that contain an
unchecked union component), so such comparisons are simply considered
erroneous.
The layout of the resulting type corresponds exactly to a C union, where
each branch of the union corresponds to a single variant in the Ada
record. The semantics of the Ada program is not changed in any way by
the pragma, i.e. provided the above restrictions are followed, and no
erroneous incorrect references to fields or erroneous comparisons occur,
the semantics is exactly as described by the Ada reference manual.
Pragma Suppress (Discriminant_Check) applies implicitly to the
type and the default convention is C
pragma Unimplemented_Unit pragma Unimplemented_Unit;
If this pragma occurs in a unit that is processed by the compiler, GNAT aborts with the message xxx not implemented, where xxx is the name of the current compilation unit. This pragma is intended to allow the compiler to handle unimplemented library units in a clean manner.
The abort only happens if code is being generated. Thus you can use specs of unimplemented packages in syntax or semantic checking mode.
pragma Unreferenced pragma Unreferenced (local_Name {, local_Name});
This pragma signals that the entities whose names are listed are deliberately not referenced. This suppresses warnings about the entities being unreferenced, and in addition a warning will be generated if one of these entities is in fact referenced.
This is particularly useful for clearly signalling that a particular parameter is not referenced in some particular subprogram implementation and that this is deliberate. It can also be useful in the case of objects declared only for their initialization or finalization side effects.
If local_Name identifies more than one matching homonym in the
current scope, then the entity most recently declared is the one to which
the pragma applies.
pragma Unreserve_All_Interrupts pragma Unreserve_All_Interrupts;
Normally certain interrupts are reserved to the implementation. Any attempt
to attach an interrupt causes Program_Error to be raised, as described in
RM C.3.2(22). A typical example is the SIGINT interrupt used in
many systems for an Ctrl-C interrupt. Normally this interrupt is
reserved to the implementation, so that Ctrl-C can be used to
interrupt execution.
If the pragma Unreserve_All_Interrupts appears anywhere in any unit in
a program, then all such interrupts are unreserved. This allows the
program to handle these interrupts, but disables their standard
functions. For example, if this pragma is used, then pressing
Ctrl-C will not automatically interrupt execution. However,
a program can then handle the SIGINT interrupt as it chooses.
For a full list of the interrupts handled in a specific implementation,
see the source code for the specification of Ada.Interrupts.Names in
file a-intnam.ads. This is a target dependent file that contains the
list of interrupts recognized for a given target. The documentation in
this file also specifies what interrupts are affected by the use of
the Unreserve_All_Interrupts pragma.
pragma Unsuppress pragma Unsuppress (IDENTIFIER [, [On =>] NAME]);
This pragma undoes the effect of a previous pragma Suppress. If
there is no corresponding pragma Suppress in effect, it has no
effect. The range of the effect is the same as for pragma
Suppress. The meaning of the arguments is identical to that used
in pragma Suppress.
One important application is to ensure that checks are on in cases where code depends on the checks for its correct functioning, so that the code will compile correctly even if the compiler switches are set to suppress checks.
pragma Use_VADS_Size pragma Use_VADS_Size;
This is a configuration pragma. In a unit to which it applies, any use of the 'Size attribute is automatically interpreted as a use of the 'VADS_Size attribute. Note that this may result in incorrect semantic processing of valid Ada 95 programs. This is intended to aid in the handling of legacy code which depends on the interpretation of Size as implemented in the VADS compiler. See description of the VADS_Size attribute for further details.
pragma Validity_Checks pragma Validity_Checks (string_LITERAL | ALL_CHECKS | On | Off);
This pragma is used in conjunction with compiler switches to control the built in validity checking provided by GNAT. The compiler switches, if set provide an initial setting for the switches, and this pragma may be used to modify these settings, or the settings may be provided entirely by the use of the pragma. This pragma can be used anywhere that a pragma is legal, including use as a configuration pragma (including use in the gnat.adc file).
The form with a string literal specifies which validity options are to be
activated. The validity checks are first set to include only the default
reference manual settings, and then a string of letters in the string
specifies the exact set of options required. The form of this string
is exactly as described for the -gnatVx compiler switch (see the
GNAT users guide for details). For example the following two methods
can be used to enable validity checking for mode in and
in out subprogram parameters:
pragma Validity_Checks ("im");
gcc -c -gnatVim ...
The form ALL_CHECKS activates all standard checks (its use is equivalent
to the use of the gnatva switch.
The forms with Off and On
can be used to temporarily disable validity checks
as shown in the following example:
pragma Validity_Checks ("c"); -- validity checks for copies
pragma Validity_Checks (Off); -- turn off validity checks
A := B; -- B will not be validity checked
pragma Validity_Checks (On); -- turn validity checks back on
A := C; -- C will be validity checked
pragma Volatile pragma Volatile (local_NAME)
This pragma is defined by the Ada 95 Reference Manual, and the GNAT implementation is fully conformant with this definition. The reason it is mentioned in this section is that a pragma of the same name was supplied in some Ada 83 compilers, including DEC Ada 83. The Ada 95 implementation of pragma Volatile is upwards compatible with the implementation in Dec Ada 83.
pragma Warnings pragma Warnings (On | Off [, LOCAL_NAME]);
Normally warnings are enabled, with the output being controlled by
the command line switch. Warnings (Off) turns off generation of
warnings until a Warnings (On) is encountered or the end of the
current unit. If generation of warnings is turned off using this
pragma, then no warning messages are output, regardless of the
setting of the command line switches.
The form with a single argument is a configuration pragma.
If the local_name parameter is present, warnings are suppressed for
the specified entity. This suppression is effective from the point where
it occurs till the end of the extended scope of the variable (similar to
the scope of Suppress).
pragma Weak_External pragma Weak_External ([Entity =>] LOCAL_NAME);
This pragma specifies that the given entity should be marked as a weak external (one that does not have to be resolved) for the linker. For further details, consult the GCC manual.
Ada 95 defines (throughout the Ada 95 reference manual, summarized in annex K), a set of attributes that provide useful additional functionality in all areas of the language. These language defined attributes are implemented in GNAT and work as described in the Ada 95 Reference Manual.
In addition, Ada 95 allows implementations to define additional attributes whose meaning is defined by the implementation. GNAT provides a number of these implementation-dependent attributes which can be used to extend and enhance the functionality of the compiler. This section of the GNAT reference manual describes these additional attributes.
Note that any program using these attributes may not be portable to other compilers (although GNAT implements this set of attributes on all platforms). Therefore if portability to other compilers is an important consideration, you should minimize the use of these attributes.
Abort_SignalStandard'Abort_Signal (Standard is the only allowed
prefix) provides the entity for the special exception used to signal
task abort or asynchronous transfer of control. Normally this attribute
should only be used in the tasking runtime (it is highly peculiar, and
completely outside the normal semantics of Ada, for a user program to
intercept the abort exception).
Address_SizeStandard'Address_Size (Standard is the only allowed
prefix) is a static constant giving the number of bits in an
Address. It is used primarily for constructing the definition of
Memory_Size in package Standard, but may be freely used in user
programs and has the advantage of being static, while a direct
reference to System.Address'Size is non-static because Address
is a private type.
Asm_InputAsm_Input attribute denotes a function that takes two
parameters. The first is a string, the second is an expression of the
type designated by the prefix. The first (string) argument is required
to be a static expression, and is the constraint for the parameter,
(e.g. what kind of register is required). The second argument is the
value to be used as the input argument. The possible values for the
constant are the same as those used in the RTL, and are dependent on
the configuration file used to built the GCC back end.
Machine Code Insertions
Asm_OutputAsm_Output attribute denotes a function that takes two
parameters. The first is a string, the second is the name of a variable
of the type designated by the attribute prefix. The first (string)
argument is required to be a static expression and designates the
constraint for the parameter (e.g. what kind of register is
required). The second argument is the variable to be updated with the
result. The possible values for constraint are the same as those used in
the RTL, and are dependent on the configuration file used to build the
GCC back end. If there are no output operands, then this argument may
either be omitted, or explicitly given as No_Output_Operands.
Machine Code Insertions
AST_EntryExtend_System (Aux_DEC)). This value enables the given entry to
be called when an AST occurs. For further details, refer to the DEC Ada
Language Reference Manual, section 9.12a.
Bit'Bit, where obj is any object, yields the bit
offset within the storage unit (byte) that contains the first bit of
storage allocated for the object. The value of this attribute is of the
type Universal_Integer, and is always a non-negative number not
exceeding the value of System.Storage_Unit.
For an object that is a variable or a constant allocated in a register, the value is zero. (The use of this attribute does not force the allocation of a variable to memory).
For an object that is a formal parameter, this attribute applies to either the matching actual parameter or to a copy of the matching actual parameter.
For an access object the value is zero. Note that
obj.all'Bit is subject to an Access_Check for the
designated object. Similarly for a record component
X.C'Bit is subject to a discriminant check and
X(I).Bit and X(I1..I2)'Bit
are subject to index checks.
This attribute is designed to be compatible with the DEC Ada 83 definition
and implementation of the Bit attribute.
Bit_Position'Bit, where R is a record object and C is one
of the fields of the record type, yields the bit
offset within the record contains the first bit of
storage allocated for the object. The value of this attribute is of the
type Universal_Integer. The value depends only on the field
C and is independent of the alignment of
the containing record R.
Code_Address'Address
attribute may be applied to subprograms in Ada 95, but the
intended effect from the Ada 95 reference manual seems to be to provide
an address value which can be used to call the subprogram by means of
an address clause as in the following example:
procedure K is ...
procedure L;
for L'Address use K'Address;
pragma Import (Ada, L);
A call to L is then expected to result in a call to K. In Ada 83, where
there were no access-to-subprogram values, this was a common work around
for getting the effect of an indirect call.
GNAT implements the above use of Address and the technique illustrated
by the example code works correctly.
However, for some purposes, it is useful to have the address of the start
of the generated code for the subprogram. On some architectures, this is
not necessarily the same as the Address value described above. For example,
the Address value may reference a subprogram descriptor rather than the
subprogram itself.
The 'Code_Address attribute, which can only be applied to
subprogram entities, always returns the address of the start of the
generated code of the specified subprogram, which may or may not be
the same value as is returned by the corresponding 'Address
attribute.
Default_Bit_OrderStandard'Default_Bit_Order (Standard is the only
permissible prefix), provides the value System.Default_Bit_Order
as a Pos value (0 for High_Order_First, 1 for
Low_Order_First). This is used to construct the definition of
Default_Bit_Order in package System.
Elaborated'Elaborated attribute must be a unit name. The
value is a Boolean which indicates whether or not the given unit has been
elaborated. This attribute is primarily intended for internal use by the
generated code for dynamic elaboration checking, but it can also be used
in user programs. The value will always be True once elaboration of all
units has been completed.
Elab_BodyElab_SpecEmaxEmax attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
Enum_Rep'Enum_Rep denotes a
function with the following specification:
function S'Enum_Rep (Arg : S'Base)
return Universal_Integer;
It is also allowable to apply Enum_Rep directly to an object of an
enumeration type or to a non-overloaded enumeration
literal. In this case S'Enum_Rep is equivalent to
typ'Enum_Rep(S) where typ is the type of the
enumeration literal or object.
The function returns the representation value for the given enumeration
value. This will be equal to value of the Pos attribute in the
absence of an enumeration representation clause. This is a static
attribute (i.e. the result is static if the argument is static).
S'Enum_Rep can also be used with integer types and objects, in which
case it simply returns the integer value. The reason for this is to allow
it to be used for (<>) discrete formal arguments in a generic unit that
can be instantiated with either enumeration types or integer types. Note
that if Enum_Rep is used on a modular type whose upper bound exceeds the
upper bound of the largest signed integer type, and the argument is a
variable, so that the universal integer calculation is done at run-time,
then the call to Enum_Rep may raise Constraint_Error.
EpsilonEpsilon attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
Fixed_Value'Fixed_Value denotes a
function with the following specification:
function S'Fixed_Value (Arg : Universal_Integer)
return S;
The value returned is the fixed-point value V such that
V = Arg * S'Small
The effect is thus equivalent to first converting the argument to the integer type used to represent S, and then doing an unchecked conversion to the fixed-point type. This attribute is primarily intended for use in implementation of the input-output functions for fixed-point values.
Has_DiscriminantsHas_Discriminants attribute is a type. The result
is a Boolean value which is True if the type has discriminants, and False
otherwise. The intended use of this attribute is in conjunction with generic
definitions. If the attribute is applied to a generic private type, it
indicates whether or not the corresponding actual type has discriminants.
ImgImg attribute differs from Image in that it may be
applied to objects as well as types, in which case it gives the
Image for the subtype of the object. This is convenient for
debugging:
Put_Line ("X = " & X'Img);
has the same meaning as the more verbose:
Put_Line ("X = " & type'Image (X));
where type is the subtype of the object X.
Integer_Value'Integer_Value denotes a
function with the following specification:
function S'Integer_Value (Arg : Universal_Fixed)
return S;
The value returned is the integer value V, such that
Arg = V * type'Small
The effect is thus equivalent to first doing an unchecked convert from the fixed-point type to its corresponding implementation type, and then converting the result to the target integer type. This attribute is primarily intended for use in implementation of the standard input-output functions for fixed-point values.
LargeLarge attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
Machine_SizeObject_Size attribute. It is
provided for compatibility with the DEC Ada 83 attribute of this name.
MantissaMantissa attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
Max_Interrupt_PriorityStandard'Max_Interrupt_Priority (Standard is the only
permissible prefix), provides the value
System.Max_Interrupt_Priority and is intended primarily for
constructing this definition in package System.
Max_PriorityStandard'Max_Priority (Standard is the only permissible
prefix) provides the value System.Max_Priority and is intended
primarily for constructing this definition in package System.
Maximum_AlignmentStandard'Maximum_Alignment (Standard is the only
permissible prefix) provides the maximum useful alignment value for the
target. This is a static value that can be used to specify the alignment
for an object, guaranteeing that it is properly aligned in all
cases. This is useful when an external object is imported and its
alignment requirements are unknown.
Mechanism_Code'Mechanism_Code yields an integer code for the
mechanism used for the result of function, and
subprogram'Mechanism_Code (n) yields the mechanism
used for formal parameter number n (a static integer value with 1
meaning the first parameter) of subprogram. The code returned is:
Values from 3 through 10 are only relevant to Digital OpenVMS implementations.
Null_Parameter'Null_Parameter denotes an imaginary object of
type or subtype T allocated at machine address zero. The attribute
is allowed only as the default expression of a formal parameter, or as
an actual expression of a subprogram call. In either case, the
subprogram must be imported.
The identity of the object is represented by the address zero in the argument list, independent of the passing mechanism (explicit or default).
This capability is needed to specify that a zero address should be
passed for a record or other composite object passed by reference.
There is no way of indicating this without the Null_Parameter
attribute.
Object_SizeNatural'Size is
31, but by default objects of type Natural will have a size of 32 bits.
Similarly, a record containing an integer and a character:
type Rec is record
I : Integer;
C : Character;
end record;
will have a size of 40 (that is Rec'Size will be 40. The
alignment will be 4, because of the
integer field, and so the default size of record objects for this type
will be 64 (8 bytes).
The type'Object_Size attribute
has been added to GNAT to allow the
default object size of a type to be easily determined. For example,
Natural'Object_Size is 32, and
Rec'Object_Size (for the record type in the above example) will be
64. Note also that, unlike the situation with the
Size attribute as defined in the Ada RM, the
Object_Size attribute can be specified individually
for different subtypes. For example:
type R is new Integer;
subtype R1 is R range 1 .. 10;
subtype R2 is R range 1 .. 10;
for R2'Object_Size use 8;
In this example, R'Object_Size and R1'Object_Size are both
32 since the default object size for a subtype is the same as the object size
for the parent subtype. This means that objects of type R
or R1 will
by default be 32 bits (four bytes). But objects of type
R2 will be only
8 bits (one byte), since R2'Object_Size has been set to 8.
Passed_By_Reference'Passed_By_Reference for any subtype type returns
a value of type Boolean value that is True if the type is
normally passed by reference and False if the type is normally
passed by copy in calls. For scalar types, the result is always False
and is static. For non-scalar types, the result is non-static.
Range_Length'Range_Length for any discrete type type yields
the number of values represented by the subtype (zero for a null
range). The result is static for static subtypes. Range_Length
applied to the index subtype of a one dimensional array always gives the
same result as Range applied to the array itself.
Safe_EmaxSafe_Emax attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
Safe_LargeSafe_Large attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
Safe_LargeSafe_Large attribute is provided for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute.
SmallSmall attribute is defined in Ada 95 only for fixed-point types.
GNAT also allows this attribute to be applied to floating-point types
for compatibility with Ada 83. See
the Ada 83 reference manual for an exact description of the semantics of
this attribute when applied to floating-point types.
Storage_UnitStandard'Storage_Unit (Standard is the only permissible
prefix) provides the value System.Storage_Unit and is intended
primarily for constructing this definition in package System.
TickStandard'Tick (Standard is the only permissible prefix)
provides the value of System.Tick and is intended primarily for
constructing this definition in package System.
To_AddressSystem'To_Address
(System is the only permissible prefix)
denotes a function identical to
System.Storage_Elements.To_Address except that
it is a static attribute. This means that if its argument is
a static expression, then the result of the attribute is a
static expression. The result is that such an expression can be
used in contexts (e.g. preelaborable packages) which require a
static expression and where the function call could not be used
(since the function call is always non-static, even if its
argument is static).
Type_Class'Type_Class for any type or subtype type yields
the value of the type class for the full type of type. If
type is a generic formal type, the value is the value for the
corresponding actual subtype. The value of this attribute is of type
System.Aux_DEC.Type_Class, which has the following definition:
type Type_Class is
(Type_Class_Enumeration,
Type_Class_Integer,
Type_Class_Fixed_Point,
Type_Class_Floating_Point,
Type_Class_Array,
Type_Class_Record,
Type_Class_Access,
Type_Class_Task,
Type_Class_Address);
Protected types yield the value Type_Class_Task, which thus
applies to all concurrent types. This attribute is designed to
be compatible with the DEC Ada 83 attribute of the same name.
UET_AddressUET_Address attribute can only be used for a prefix which
denotes a library package. It yields the address of the unit exception
table when zero cost exception handling is used. This attribute is
intended only for use within the GNAT implementation. See the unit
Ada.Exceptions in files a-except.ads and a-except.adb
for details on how this attribute is used in the implementation.
Universal_Literal_StringUniversal_Literal_String must be a named
number. The static result is the string consisting of the characters of
the number as defined in the original source. This allows the user
program to access the actual text of named numbers without intermediate
conversions and without the need to enclose the strings in quotes (which
would preclude their use as numbers). This is used internally for the
construction of values of the floating-point attributes from the file
ttypef.ads, but may also be used by user programs.
Unrestricted_AccessUnrestricted_Access attribute is similar to Access
except that all accessibility and aliased view checks are omitted. This
is a user-beware attribute. It is similar to
Address, for which it is a desirable replacement where the value
desired is an access type. In other words, its effect is identical to
first applying the Address attribute and then doing an unchecked
conversion to a desired access type. In GNAT, but not necessarily in
other implementations, the use of static chains for inner level
subprograms means that Unrestricted_Access applied to a
subprogram yields a value that can be called as long as the subprogram
is in scope (normal Ada 95 accessibility rules restrict this usage).
VADS_Size'VADS_Size attribute is intended to make it easier to port
legacy code which relies on the semantics of 'Size as implemented
by the VADS Ada 83 compiler. GNAT makes a best effort at duplicating the
same semantic interpretation. In particular, 'VADS_Size applied
to a predefined or other primitive type with no Size clause yields the
Object_Size (for example, Natural'Size is 32 rather than 31 on
typical machines). In addition 'VADS_Size applied to an object
gives the result that would be obtained by applying the attribute to
the corresponding type.
Value_Size'Value_Size is the number of bits required to represent
a value of the given subtype. It is the same as type'Size,
but, unlike Size, may be set for non-first subtypes.
Wchar_T_SizeStandard'Wchar_T_Size (Standard is the only permissible
prefix) provides the size in bits of the C wchar_t type
primarily for constructing the definition of this type in
package Interfaces.C.
Word_SizeStandard'Word_Size (Standard is the only permissible
prefix) provides the value System.Word_Size and is intended
primarily for constructing this definition in package System.
The main text of the Ada 95 Reference Manual describes the required behavior of all Ada 95 compilers, and the GNAT compiler conforms to these requirements.
In addition, there are sections throughout the Ada 95 reference manual headed by the phrase “implementation advice”. These sections are not normative, i.e. they do not specify requirements that all compilers must follow. Rather they provide advice on generally desirable behavior. You may wonder why they are not requirements. The most typical answer is that they describe behavior that seems generally desirable, but cannot be provided on all systems, or which may be undesirable on some systems.
As far as practical, GNAT follows the implementation advice sections in the Ada 95 Reference Manual. This chapter contains a table giving the reference manual section number, paragraph number and several keywords for each advice. Each entry consists of the text of the advice followed by the GNAT interpretation of this advice. Most often, this simply says “followed”, which means that GNAT follows the advice. However, in a number of cases, GNAT deliberately deviates from this advice, in which case the text describes what GNAT does and why.
If an implementation detects the use of an unsupported Specialized Needs
Annex feature at run time, it should raise Program_Error if
feasible.
|
| If an implementation wishes to provide implementation-defined extensions to the functionality of a language-defined library unit, it should normally do so by adding children to the library unit. |
If an implementation detects a bounded error or erroneous
execution, it should raise Program_Error.
|
| Normally, implementation-defined pragmas should have no semantic effect for error-free programs; that is, if the implementation-defined pragmas are removed from a working program, the program should still be legal, and should still have the same semantics. |
Abort_DeferAda_83AssertCPP_ClassCPP_ConstructorCPP_VirtualCPP_VtableDebugInterface_NameMachine_AttributeUnimplemented_UnitUnchecked_UnionIn each of the above cases, it is essential to the purpose of the pragma
that this advice not be followed. For details see the separate section
on implementation defined pragmas.
| Normally, an implementation should not define pragmas that can make an illegal program legal, except as follows: |
A pragma used to complete a declaration, such as a pragma Import;
|
A pragma used to configure the environment by adding, removing, or
replacing library_items.
|
If an implementation supports a mode with alternative interpretations
for Character and Wide_Character, the set of graphic
characters of Character should nevertheless remain a proper
subset of the set of graphic characters of Wide_Character. Any
character set “localizations” should be reflected in the results of
the subprograms defined in the language-defined package
Characters.Handling (see A.3) available in such a mode. In a mode with
an alternative interpretation of Character, the implementation should
also support a corresponding change in what is a legal
identifier_letter.
|
An implementation should support Long_Integer in addition to
Integer if the target machine supports 32-bit (or longer)
arithmetic. No other named integer subtypes are recommended for package
Standard. Instead, appropriate named integer subtypes should be
provided in the library package Interfaces (see B.2).
|
Long_Integer is supported. Other standard integer types are supported
so this advice is not fully followed. These types
are supported for convenient interface to C, and so that all hardware
types of the machine are easily available.
An implementation for a two's complement machine should support
modular types with a binary modulus up to System.Max_Int*2+2. An
implementation should support a non-binary modules up to Integer'Last.
|
For the evaluation of a call on S'Pos for an enumeration
subtype, if the value of the operand does not correspond to the internal
code for any enumeration literal of its type (perhaps due to an
un-initialized variable), then the implementation should raise
Program_Error. This is particularly important for enumeration
types with noncontiguous internal codes specified by an
enumeration_representation_clause.
|
An implementation should support Long_Float in addition to
Float if the target machine supports 11 or more digits of
precision. No other named floating point subtypes are recommended for
package Standard. Instead, appropriate named floating point subtypes
should be provided in the library package Interfaces (see B.2).
|
Short_Float and Long_Long_Float are also provided. The
former provides improved compatibility with other implementations
supporting this type. The latter corresponds to the highest precision
floating-point type supported by the hardware. On most machines, this
will be the same as Long_Float, but on some machines, it will
correspond to the IEEE extended form. The notable case is all ia32
(x86) implementations, where Long_Long_Float corresponds to
the 80-bit extended precision format supported in hardware on this
processor. Note that the 128-bit format on SPARC is not supported,
since this is a software rather than a hardware format.
An implementation should normally represent multidimensional arrays in
row-major order, consistent with the notation used for multidimensional
array aggregates (see 4.3.3). However, if a pragma Convention
(Fortran, ...) applies to a multidimensional array type, then
column-major order should be used instead (see B.5, “Interfacing with
Fortran”).
|
Whenever possible in an implementation, the value of Duration'Small
should be no greater than 100 microseconds.
|
Duration'Small = 10**(−9)).
The time base for delay_relative_statements should be monotonic;
it need not be the same time base as used for Calendar.Clock.
|
| In an implementation, a type declared in a pre-elaborated package should have the same representation in every elaboration of a given version of the package, whether the elaborations occur in distinct executions of the same program, or in executions of distinct programs or partitions that include the given version. |
Exception_Message by default and Exception_Information
should produce information useful for
debugging. Exception_Message should be short, about one
line. Exception_Information can be long. Exception_Message
should not include the
Exception_Name. Exception_Information should include both
the Exception_Name and the Exception_Message.
|
Exception_Message, the compiler generates one containing the location
of the raise statement. This location has the form “file:line”, where
file is the short file name (without path information) and line is the line
number in the file. Note that in the case of the Zero Cost Exception
mechanism, these messages become redundant with the Exception_Information that
contains a full backtrace of the calling sequence, so they are disabled.
To disable explicitly the generation of the source location message, use the
Pragma Discard_Names.
| The implementation should minimize the code executed for checks that have been suppressed. |
| The recommended level of support for all representation items is qualified as follows: |
| An implementation need not support representation items containing non-static expressions, except that an implementation should support a representation item for a given entity if each non-static expression in the representation item is a name that statically denotes a constant declared before the entity. |
X : typ;
for X'Address use To_address (16#2000#);
will be rejected, since the To_Address expression is non-static. Instead write:
X_Address : constant Address : =
To_Address ((16#2000#);
X : typ;
for X'Address use X_Address;
An implementation need not support a specification for the Size
for a given composite subtype, nor the size or storage place for an
object (including a component) of a given composite subtype, unless the
constraints on the subtype and its composite subcomponents (if any) are
all static constraints.
|
| An aliased component, or a component whose type is by-reference, should always be allocated at an addressable location. |
| If a type is packed, then the implementation should try to minimize storage allocated to objects of the type, possibly at the expense of speed of accessing components, subject to reasonable complexity in addressing calculations. |
The recommended level of support pragma Pack is:
For a packed record type, the components should be packed as tightly as
possible subject to the Sizes of the component subtypes, and subject to
any |
| An implementation should support Address clauses for imported subprograms. |
For an array X, X'Address should point at the first
component of the array, and not at the array bounds.
|
The recommended level of support for the Address attribute is:
X |
An implementation should support Address clauses for imported
subprograms.
|
| Objects (including subcomponents) that are aliased or of a by-reference type should be allocated on storage element boundaries. |
If the Address of an object is specified, or it is imported or exported,
then the implementation should not perform optimizations based on
assumptions of no aliases.
|
The recommended level of support for the Alignment attribute for
subtypes is:
An implementation should support specified Alignments that are factors and multiples of the number of storage elements per word, subject to the following: |
An implementation need not support specified Alignments for
combinations of Sizes and Alignments that cannot be easily
loaded and stored by available machine instructions.
|
An implementation need not support specified Alignments that are
greater than the maximum Alignment the implementation ever returns by
default.
|
The recommended level of support for the Alignment attribute for
objects is:
Same as above, for subtypes, but in addition: |
For stand-alone library-level objects of statically constrained
subtypes, the implementation should support all Alignments
supported by the target linker. For example, page alignment is likely to
be supported for such objects, but not for subtypes.
|
The recommended level of support for the Size attribute of
objects is:
A |
If the Size of a subtype is specified, and allows for efficient
independent addressability (see 9.10) on the target architecture, then
the Size of the following objects of the subtype should equal the
Size of the subtype:
Aliased objects (including components). |
Size clause on a composite subtype should not affect the
internal layout of components.
|
The recommended level of support for the Size attribute of subtypes is:
|
The Size (if not specified) of a static discrete or fixed point
subtype should be the number of bits needed to represent each value
belonging to the subtype using an unbiased representation, leaving space
for a sign bit only if the subtype contains negative values. If such a
subtype is a first subtype, then an implementation should support a
specified Size for it that reflects this representation.
|
For a subtype implemented with levels of indirection, the Size
should include the size of the pointers, but not the size of what they
point at.
|
The recommended level of support for the Component_Size
attribute is:
|
An implementation need not support specified Component_Sizes that are
less than the Size of the component subtype.
|
An implementation should support specified Component_Sizes that
are factors and multiples of the word size. For such
Component_Sizes, the array should contain no gaps between
components. For other Component_Sizes (if supported), the array
should contain no gaps between components when packing is also
specified; the implementation should forbid this combination in cases
where it cannot support a no-gaps representation.
|
|
The recommended level of support for enumeration representation clauses
is:
An implementation need not support enumeration representation clauses
for boolean types, but should at minimum support the internal codes in
the range |
The recommended level of support for
record_representation_clauses is:
An implementation should support storage places that can be extracted with a load, mask, shift sequence of machine code, and set with a load, shift, mask, store sequence, given the available machine instructions and run-time model. |
A storage place should be supported if its size is equal to the
Size of the component subtype, and it starts and ends on a
boundary that obeys the Alignment of the component subtype.
|
If the default bit ordering applies to the declaration of a given type,
then for a component whose subtype's Size is less than the word
size, any storage place that does not cross an aligned word boundary
should be supported.
|
| An implementation may reserve a storage place for the tag field of a tagged type, and disallow other components from overlapping that place. |
An implementation need not support a component_clause for a
component of an extension part if the storage place is not after the
storage places of all components of the parent type, whether or not
those storage places had been specified.
|
| If a component is represented using some form of pointer (such as an offset) to the actual data of the component, and this data is contiguous with the rest of the object, then the storage place attributes should reflect the place of the actual data, not the pointer. If a component is allocated discontinuously from the rest of the object, then a warning should be generated upon reference to one of its storage place attributes. |
| The recommended level of support for the non-default bit ordering is: |
If Word_Size = Storage_Unit, then the implementation
should support the non-default bit ordering in addition to the default
bit ordering.
|
Address should be of a private type.
|
Operations in System and its children should reflect the target
environment semantics as closely as is reasonable. For example, on most
machines, it makes sense for address arithmetic to “wrap around”.
Operations that do not make sense should raise Program_Error.
|
Program_Error, since all operations make sense.
The Size of an array object should not include its bounds; hence,
the bounds should not be part of the converted data.
|
| The implementation should not generate unnecessary run-time checks to ensure that the representation of S is a representation of the target type. It should take advantage of the permission to return by reference when possible. Restrictions on unchecked conversions should be avoided unless required by the target environment. |
| The recommended level of support for unchecked conversions is: |
| Unchecked conversions should be supported and should be reversible in the cases where this clause defines the result. To enable meaningful use of unchecked conversion, a contiguous representation should be used for elementary subtypes, for statically constrained array subtypes whose component subtype is one of the subtypes described in this paragraph, and for record subtypes without discriminants whose component subtypes are described in this paragraph. |
| An implementation should document any cases in which it dynamically allocates heap storage for a purpose other than the evaluation of an allocator. |
| A default (implementation-provided) storage pool for an access-to-constant type should not have overhead to support deallocation of individual objects. |
| A storage pool for an anonymous access type should be created at the point of an allocator for the type, and be reclaimed when the designated object becomes inaccessible. |
For a standard storage pool, Free should actually reclaim the
storage.
|
If a stream element is the same size as a storage element, then the
normal in-memory representation should be used by Read and
Write for scalar objects. Otherwise, Read and Write
should use the smallest number of stream elements needed to represent
all values in the base range of the scalar type.
|
| If an implementation provides additional named predefined integer types, then the names should end with Integer as in Long_Integer. If an implementation provides additional named predefined floating point types, then the names should end with Float as in Long_Float. |
Ada.Characters.Handling
If an implementation provides a localized definition of Character
or Wide_Character, then the effects of the subprograms in
Characters.Handling should reflect the localizations. See also
3.5.2.
|
| Bounded string objects should not be implemented by implicit pointers and dynamic allocation. |
Any storage associated with an object of type Generator should be
reclaimed on exit from the scope of the object.
|
If the generator period is sufficiently long in relation to the number
of distinct initiator values, then each possible value of
Initiator passed to Reset should initiate a sequence of
random numbers that does not, in a practical sense, overlap the sequence
initiated by any other value. If this is not possible, then the mapping
between initiator values and generator states should be a rapidly
varying function of the initiator value.
|
Get_Immediate
The Get_Immediate procedures should be implemented with
unbuffered input. For a device such as a keyboard, input should be
available if a key has already been typed, whereas for a disk
file, input should always be available except at end of file. For a file
associated with a keyboard-like device, any line-editing features of the
underlying operating system should be disabled during the execution of
Get_Immediate.
|
Export
If an implementation supports pragma Export to a given language,
then it should also allow the main subprogram to be written in that
language. It should support some mechanism for invoking the elaboration
of the Ada library units included in the system, and for invoking the
finalization of the environment task. On typical systems, the
recommended mechanism is to provide two subprograms whose link names are
adainit and adafinal. adainit should contain the
elaboration code for library units. adafinal should contain the
finalization code. These subprograms should have no effect the second
and subsequent time they are called.
|
Automatic elaboration of pre-elaborated packages should be
provided when pragma Export is supported.
|
adainit must be called to elaborate pre-elaborated
packages.
For each supported convention L other than Intrinsic, an
implementation should support Import and Export pragmas
for objects of L-compatible types and for subprograms, and pragma
Convention for L-eligible types and for subprograms,
presuming the other language has corresponding features. Pragma
Convention need not be supported for scalar types.
|
Interfaces
For each implementation-defined convention identifier, there should be a
child package of package Interfaces with the corresponding name. This
package should contain any declarations that would be useful for
interfacing to the language (implementation) represented by the
convention. Any declarations useful for interfacing to any language on
the given hardware architecture should be provided directly in
Interfaces.
|
Interfaces.CPP, used
for interfacing to C++.
| An implementation supporting an interface to C, COBOL, or Fortran should provide the corresponding package or packages described in the following clauses. |
| An implementation should support the following interface correspondences between Ada and C. |
| An Ada procedure corresponds to a void-returning C function. |
| An Ada function corresponds to a non-void C function. |
An Ada in scalar parameter is passed as a scalar argument to a C
function.
|
An Ada in parameter of an access-to-object type with designated
type T is passed as a t* argument to a C function,
where t is the C type corresponding to the Ada type T.
|
An Ada access T parameter, or an Ada out or in out
parameter of an elementary type T, is passed as a t*
argument to a C function, where t is the C type corresponding to
the Ada type T. In the case of an elementary out or
in out parameter, a pointer to a temporary copy is used to
preserve by-copy semantics.
|
An Ada parameter of a record type T, of any mode, is passed as a
t* argument to a C function, where t is the C
structure corresponding to the Ada type T.
|
An Ada parameter of an array type with component type T, of any
mode, is passed as a t* argument to a C function, where
t is the C type corresponding to the Ada type T.
|
| An Ada parameter of an access-to-subprogram type is passed as a pointer to a C function whose prototype corresponds to the designated subprogram's specification. |
| An Ada implementation should support the following interface correspondences between Ada and COBOL. |
| An Ada access T parameter is passed as a BY REFERENCE data item of the COBOL type corresponding to T. |
| An Ada in scalar parameter is passed as a BY CONTENT data item of the corresponding COBOL type. |
| Any other Ada parameter is passed as a BY REFERENCE data item of the COBOL type corresponding to the Ada parameter type; for scalars, a local copy is used if necessary to ensure by-copy semantics. |
| An Ada implementation should support the following interface correspondences between Ada and Fortran: |
| An Ada procedure corresponds to a Fortran subroutine. |
| An Ada function corresponds to a Fortran function. |
| An Ada parameter of an elementary, array, or record type T is passed as a T argument to a Fortran procedure, where T is the Fortran type corresponding to the Ada type T, and where the INTENT attribute of the corresponding dummy argument matches the Ada formal parameter mode; the Fortran implementation's parameter passing conventions are used. For elementary types, a local copy is used if necessary to ensure by-copy semantics. |
| An Ada parameter of an access-to-subprogram type is passed as a reference to a Fortran procedure whose interface corresponds to the designated subprogram's specification. |
| The machine code or intrinsic support should allow access to all operations normally available to assembly language programmers for the target environment, including privileged instructions, if any. |
The interfacing pragmas (see Annex B) should support interface to
assembler; the default assembler should be associated with the
convention identifier Assembler.
|
| If an entity is exported to assembly language, then the implementation should allocate it at an addressable location, and should ensure that it is retained by the linking process, even if not otherwise referenced from the Ada code. The implementation should assume that any call to a machine code or assembler subprogram is allowed to read or update every object that is specified as exported. |
| The implementation should ensure that little or no overhead is associated with calling intrinsic and machine-code subprograms. |
| It is recommended that intrinsic subprograms be provided for convenient access to any machine operations that provide special capabilities or efficiency and that are not otherwise available through the language constructs. |
| Atomic read-modify-write operations—e.g., test and set, compare and swap, decrement and test, enqueue/dequeue. |
| Standard numeric functions—e.g., sin, log. |
| String manipulation operations—e.g., translate and test. |
| Vector operations—e.g., compare vector against thresholds. |
| Direct operations on I/O ports. |
If the Ceiling_Locking policy is not in effect, the
implementation should provide means for the application to specify which
interrupts are to be blocked during protected actions, if the underlying
system allows for a finer-grain control of interrupt blocking.
|
| Whenever possible, the implementation should allow interrupt handlers to be called directly by the hardware. |
| Whenever practical, violations of any implementation-defined restrictions should be detected before run time. |
Interrupts
If implementation-defined forms of interrupt handler procedures are
supported, such as protected procedures with parameters, then for each
such form of a handler, a type analogous to Parameterless_Handler
should be specified in a child package of Interrupts, with the
same operations as in the predefined package Interrupts.
|
| It is recommended that pre-elaborated packages be implemented in such a way that there should be little or no code executed at run time for the elaboration of entities not already covered by the Implementation Requirements. |
Discard_Names
| If the pragma applies to an entity, then the implementation should reduce the amount of storage used for storing names associated with that entity. |
| Some implementations are targeted to domains in which memory use at run time must be completely deterministic. For such implementations, it is recommended that the storage for task attributes will be pre-allocated statically and not from the heap. This can be accomplished by either placing restrictions on the number and the size of the task's attributes, or by using the pre-allocated storage for the first N attribute objects, and the heap for the others. In the latter case, N should be documented. |
| The implementation should use names that end with _Locking for locking policies defined by the implementation. |
Inheritance_Locking) follows this suggestion.
| Names that end with _Queuing should be used for all implementation-defined queuing policies. |
Even though the abort_statement is included in the list of
potentially blocking operations (see 9.5.1), it is recommended that this
statement be implemented in a way that never requires the task executing
the abort_statement to block.
|
| On a multi-processor, the delay associated with aborting a task on another processor should be bounded; the implementation should use periodic polling, if necessary, to achieve this. |
| When feasible, the implementation should take advantage of the specified restrictions to produce a more efficient implementation. |
Ravenscar and pragma
Restricted_Run_Time for more details.
When appropriate, implementations should provide configuration
mechanisms to change the value of Tick.
|
It is recommended that Calendar.Clock and Real_Time.Clock
be implemented as transformations of the same time base.
|
It is recommended that the best time base which exists in
the underlying system be available to the application through
Clock. Best may mean highest accuracy or largest range.
|
| Whenever possible, the PCS on the called partition should allow for multiple tasks to call the RPC-receiver with different messages and should allow them to block until the corresponding subprogram body returns. |
The Write operation on a stream of type Params_Stream_Type
should raise Storage_Error if it runs out of space trying to
write the Item into the stream.
|
If COBOL (respectively, C) is widely supported in the target
environment, implementations supporting the Information Systems Annex
should provide the child package Interfaces.COBOL (respectively,
Interfaces.C) specified in Annex B and should support a
convention_identifier of COBOL (respectively, C) in the interfacing
pragmas (see Annex B), thus allowing Ada programs to interface with
programs written in that language.
|
| Packed decimal should be used as the internal representation for objects of subtype S when S'Machine_Radix = 10. |
If Fortran (respectively, C) is widely supported in the target
environment, implementations supporting the Numerics Annex
should provide the child package Interfaces.Fortran (respectively,
Interfaces.C) specified in Annex B and should support a
convention_identifier of Fortran (respectively, C) in the interfacing
pragmas (see Annex B), thus allowing Ada programs to interface with
programs written in that language.
|
| Because the usual mathematical meaning of multiplication of a complex operand and a real operand is that of the scaling of both components of the former by the latter, an implementation should not perform this operation by first promoting the real operand to complex type and then performing a full complex multiplication. In systems that, in the future, support an Ada binding to IEC 559:1989, the latter technique will not generate the required result when one of the components of the complex operand is infinite. (Explicit multiplication of the infinite component by the zero component obtained during promotion yields a NaN that propagates into the final result.) Analogous advice applies in the case of multiplication of a complex operand and a pure-imaginary operand, and in the case of division of a complex operand by a real or pure-imaginary operand. |
Similarly, because the usual mathematical meaning of addition of a
complex operand and a real operand is that the imaginary operand remains
unchanged, an implementation should not perform this operation by first
promoting the real operand to complex type and then performing a full
complex addition. In implementations in which the Signed_Zeros
attribute of the component type is True (and which therefore
conform to IEC 559:1989 in regard to the handling of the sign of zero in
predefined arithmetic operations), the latter technique will not
generate the required result when the imaginary component of the complex
operand is a negatively signed zero. (Explicit addition of the negative
zero to the zero obtained during promotion yields a positive zero.)
Analogous advice applies in the case of addition of a complex operand
and a pure-imaginary operand, and in the case of subtraction of a
complex operand and a real or pure-imaginary operand.
|
Implementations in which Real'Signed_Zeros is True should
attempt to provide a rational treatment of the signs of zero results and
result components. As one example, the result of the Argument
function should have the sign of the imaginary component of the
parameter X when the point represented by that parameter lies on
the positive real axis; as another, the sign of the imaginary component
of the Compose_From_Polar function should be the same as
(respectively, the opposite of) that of the Argument parameter when that
parameter has a value of zero and the Modulus parameter has a
nonnegative (respectively, negative) value.
|
Implementations in which Complex_Types.Real'Signed_Zeros is
True should attempt to provide a rational treatment of the signs
of zero results and result components. For example, many of the complex
elementary functions have components that are odd functions of one of
the parameter components; in these cases, the result component should
have the sign of the parameter component at the origin. Other complex
elementary functions have zero components whose sign is opposite that of
a parameter component at the origin, or is always positive or always
negative.
|
The versions of the forward trigonometric functions without a
Cycle parameter should not be implemented by calling the
corresponding version with a Cycle parameter of
2.0*Numerics.Pi, since this will not provide the required
accuracy in some portions of the domain. For the same reason, the
version of Log without a Base parameter should not be
implemented by calling the corresponding version with a Base
parameter of Numerics.e.
|
The version of the Compose_From_Polar function without a
Cycle parameter should not be implemented by calling the
corresponding version with a Cycle parameter of
2.0*Numerics.Pi, since this will not provide the required
accuracy in some portions of the domain.
|
In addition to the implementation dependent pragmas and attributes, and the implementation advice, there are a number of other features of Ada 95 that are potentially implementation dependent. These are mentioned throughout the Ada 95 Reference Manual, and are summarized in annex M.
A requirement for conforming Ada compilers is that they provide documentation describing how the implementation deals with each of these issues. In this chapter, you will find each point in annex M listed followed by a description in italic font of how GNAT handles the implementation dependence.
You can use this chapter as a guide to minimizing implementation dependent features in your programs if portability to other compilers and other operating systems is an important consideration. The numbers in each section below correspond to the paragraph number in the Ada 95 Reference Manual.
| 2. Whether or not each recommendation given in Implementation Advice is followed. See 1.1.2(37). |
| 3. Capacity limitations of the implementation. See 1.1.3(3). |
| 4. Variations from the standard that are impractical to avoid given the implementation's execution environment. See 1.1.3(6). |
5. Which code_statements cause external
interactions. See 1.1.3(10).
|
code_statement can potentially cause external interactions.
| 6. The coded representation for the text of an Ada program. See 2.1(4). |
| 7. The control functions allowed in comments. See 2.1(14). |
| 8. The representation for an end of line. See 2.2(2). |
| 9. Maximum supported line length and lexical element length. See 2.2(15). |
| 10. Implementation defined pragmas. See 2.8(14). |
11. Effect of pragma Optimize. See 2.8(27).
|
Optimize, if given with a Time or Space
parameter, checks that the optimization flag is set, and aborts if it is
not.
12. The sequence of characters of the value returned by
S'Image when some of the graphic characters of
S'Wide_Image are not defined in Character. See
3.5(37).
|
13. The predefined integer types declared in
Standard. See 3.5.4(25).
|
Short_Short_IntegerShort_IntegerIntegerLong_IntegerLong_Long_Integer| 14. Any nonstandard integer types and the operators defined for them. See 3.5.4(26). |
| 15. Any nonstandard real types and the operators defined for them. See 3.5.6(8). |
| 16. What combinations of requested decimal precision and range are supported for floating point types. See 3.5.7(7). |
17. The predefined floating point types declared in
Standard. See 3.5.7(16).
|
Short_FloatFloatLong_FloatLong_Long_Float| 18. The small of an ordinary fixed point type. See 3.5.9(8). |
Fine_Delta is 2**(−63)
| 19. What combinations of small, range, and digits are supported for fixed point types. See 3.5.9(10). |
Fine_Delta and do not result in a mantissa larger than 63 bits.
If the mantissa is larger than 53 bits on machines where Long_Long_Float
is 64 bits (true of all architectures except ia32), then the output from
Text_IO is accurate to only 53 bits, rather than the full mantissa. This
is because floating-point conversions are used to convert fixed point.
20. The result of Tags.Expanded_Name for types declared
within an unnamed block_statement. See 3.9(10).
|
Bnnn, where nnn is a
decimal integer are allocated.
| 21. Implementation-defined attributes. See 4.1.4(12). |
| 22. Any implementation-defined time types. See 9.6(6). |
| 23. The time base associated with relative delays. |
gettimeofday.
24. The time base of the type Calendar.Time. See
9.6(23).
|
gettimeofday.
25. The time zone used for package Calendar
operations. See 9.6(24).
|
Calendar is the current system time zone
setting for local time, as accessed by the C library function
localtime.
26. Any limit on delay_until_statements of
select_statements. See 9.6(29).
|
27. Whether or not two non overlapping parts of a composite
object are independently addressable, in the case where packing, record
layout, or Component_Size is specified for the object. See
9.10(1).
|
| 28. The representation for a compilation. See 10.1(2). |
gcc command.
| 29. Any restrictions on compilations that contain multiple compilation_units. See 10.1(4). |
| 30. The mechanisms for creating an environment and for adding and replacing compilation units. See 10.1.4(3). |
| 31. The manner of explicitly assigning library units to a partition. See 10.2(2). |
If the partition contains no main program, or if the main program is in
a language other than Ada, then GNAT
provides the binder options -z and -n respectively, and in this case a
list of units can be explicitly supplied to the binder for inclusion in
the partition (all units needed by these units will also be included
automatically). For full details on the use of these options, refer to
the GNAT User's Guide sections on Binding and Linking.
| 32. The implementation-defined means, if any, of specifying which compilation units are needed by a given compilation unit. See 10.2(2). |
| 33. The manner of designating the main subprogram of a partition. See 10.2(7). |
34. The order of elaboration of library_items. See
10.2(18).
|
| 35. Parameter passing and function return for the main subprogram. See 10.2(21). |
| 36. The mechanisms for building and running partitions. See 10.2(24). |
| 37. The details of program execution, including program termination. See 10.2(25). |
| 38. The semantics of any non-active partitions supported by the implementation. See 10.2(28). |
39. The information returned by Exception_Message. See
11.4.1(10).
|
40. The result of Exceptions.Exception_Name for types
declared within an unnamed block_statement. See 11.4.1(12).
|
Bnnn
where nnn is an integer.
41. The information returned by
Exception_Information. See 11.4.1(13).
|
Exception_Information returns a string in the following format:
Exception_Name: nnnnn
Message: mmmmm
PID: ppp
Call stack traceback locations:
0xhhhh 0xhhhh 0xhhhh ... 0xhhh
where
nnnn is the fully qualified name of the exception in all upper
case letters. This line is always present.
mmmm is the message (this line present only if message is non-null)
ppp is the Process Id value as a decimal integer (this line is
present only if the Process Id is non-zero). Currently we are
not making use of this field.
The line terminator sequence at the end of each line, including
the last line is a single LF character (16#0A#).
| 42. Implementation-defined check names. See 11.5(27). |
| 43. The interpretation of each aspect of representation. See 13.1(20). |
| 44. Any restrictions placed upon representation items. See 13.1(20). |
45. The meaning of Size for indefinite subtypes. See
13.3(48).
|
| 46. The default external representation for a type tag. See 13.3(75). |
| 47. What determines whether a compilation unit is the same in two different partitions. See 13.3(76). |
| 48. Implementation-defined components. See 13.5.1(15). |
49. If Word_Size = Storage_Unit, the default bit
ordering. See 13.5.3(5).
|
Word_Size (32) is not the same as Storage_Unit (8) for this
implementation, so no non-default bit ordering is supported. The default
bit ordering corresponds to the natural endianness of the target architecture.
50. The contents of the visible part of package System
and its language-defined children. See 13.7(2).
|
51. The contents of the visible part of package
System.Machine_Code, and the meaning of
code_statements. See 13.8(7).
|
| 52. The effect of unchecked conversion. See 13.9(11). |
53. The manner of choosing a storage pool for an access type
when Storage_Pool is not specified for the type. See 13.11(17).
|
Storage_Pool is not specified depending whether the type is local
to a subprogram or defined at the library level and whether
Storage_Sizeis specified or not. See documentation in the runtime
library units System.Pool_Global, System.Pool_Size and
System.Pool_Local in files s-poosiz.ads,
s-pooglo.ads and s-pooloc.ads for full details on the
default pools used.
| 54. Whether or not the implementation provides user-accessible names for the standard pool type(s). See 13.11(17). |
with'ing
these units.
55. The meaning of Storage_Size. See 13.11(18).
|
Storage_Size is measured in storage units, and refers to the
total space available for an access type collection, or to the primary
stack space for a task.
| 56. Implementation-defined aspects of storage pools. See 13.11(22). |
57. The set of restrictions allowed in a pragma
Restrictions. See 13.12(7).
|
Boolean_Entry_BarriersRavenscar).
Max_Entry_Queue_Depth => ExprNo_CalendarAda.Calendar.
No_Dynamic_InterruptsNo_Enumeration_MapsNo_Entry_Calls_In_Elaboration_CodeNo_Exception_HandlersNo_Implicit_ConditionalsNo_Implicit_Loopsfor loops, either by modifying
the generated code where possible,
or by rejecting any construct that would otherwise generate an implicit
for loop. The details and use of this restriction are described in
more detail in the High Integrity product documentation.
No_Local_Protected_ObjectsNo_Protected_Type_AllocatorsNo_Secondary_StackNo_Select_Statementsselect may not appear.
This is one of the restrictions of the Ravenscar
profile for limited tasking (see also pragma Ravenscar).
No_Standard_Storage_PoolsNo_StreamsAda.Streams.
No_Task_AttributesAda.Task_Attributes.
No_Task_TerminationNo_TaskingMax_Tasks => 0
except that violations are caught at compile time and cause an error message
to be output either by the compiler or binder.
No_Wide_CharactersWide_Character or Wide_String
appear, and that no wide character literals
appear in the program (that is literals representing characters not in
type Character.
Static_PrioritiesAda.Dynamic_Priorities.
Static_Storage_SizeThe second set of implementation dependent restriction identifiers does not require partition-wide consistency. The restriction may be enforced for a single compilation unit without any effect on any of the other compilation units in the partition.
No_Elaboration_CodePreelaborate. There are cases in which pragma Preelaborate
still permits code to be generated (e.g. code to initialize a large
array to all zeroes), and there are cases of units which do not meet
the requirements for pragma Preelaborate, but for which no elaboration
code is generated. Generally, it is the case that preelaborable units
will meet the restrictions, with the exception of large aggregates
initialized with an others_clause, and exception declarations (which
generate calls to a run-time registry procedure). Note that this restriction
is enforced on a unit by unit basis, it need not be obeyed consistently
throughout a partition.
No_Entry_QueueNo_Implementation_AttributesNo_Implementation_PragmasNo_Implementation_RestrictionsNo_Implementation_Restrictions itself)
are present. With this restriction, the only other restriction identifiers
that can be used are those defined in the Ada 95 Reference Manual.
58. The consequences of violating limitations on
Restrictions pragmas. See 13.12(9).
|
59. The representation used by the Read and
Write attributes of elementary types in terms of stream
elements. See 13.13.2(9).
|
'Size value, and the natural ordering of the machine.
60. The names and characteristics of the numeric subtypes
declared in the visible part of package Standard. See A.1(3).
|
| 61. The accuracy actually achieved by the elementary functions. See A.5.1(1). |
62. The sign of a zero result from some of the operators or
functions in Numerics.Generic_Elementary_Functions, when
Float_Type'Signed_Zeros is True. See A.5.1(46).
|
63. The value of
Numerics.Float_Random.Max_Image_Width. See A.5.2(27).
|
64. The value of
Numerics.Discrete_Random.Max_Image_Width. See A.5.2(27).
|
| 65. The algorithms for random number generation. See A.5.2(32). |
| 66. The string representation of a random number generator's state. See A.5.2(38). |
| 67. The minimum time interval between calls to the time-dependent Reset procedure that are guaranteed to initiate different random number sequences. See A.5.2(45). |
68. The values of the Model_Mantissa,
Model_Emin, Model_Epsilon, Model,
Safe_First, and Safe_Last attributes, if the Numerics
Annex is not supported. See A.5.3(72).
|
| 69. Any implementation-defined characteristics of the input-output packages. See A.7(14). |
70. The value of Buffer_Size in Storage_IO. See
A.9(10).
|
Buffer_Size is
the value of type'Size rounded up to the next storage unit
boundary.
| 71. External files for standard input, standard output, and standard error See A.10(5). |
72. The accuracy of the value produced by Put. See
A.10.9(36).
|
73. The meaning of Argument_Count, Argument, and
Command_Name. See A.15(1).
|
argv and argc parameters of the
main program in the natural manner.
| 74. Implementation-defined convention names. See B.1(11). |
AdaAssemblerAsmAssemblyCC_Pass_By_CopyCOBOLCPPDefaultExternalFortranIntrinsicImport with convention Intrinsic, see
separate section on Intrinsic Subprograms.
StdcallDLLWin32StubbedProgram_Error exception. If a
pragma Import specifies convention stubbed then no body need
be present at all. This convention is useful during development for the
inclusion of subprograms whose body has not yet been written.
| 75. The meaning of link names. See B.1(36). |
| 76. The manner of choosing link names when neither the link name nor the address of an imported or exported entity is specified. See B.1(36). |
77. The effect of pragma Linker_Options. See B.1(37).
|
Linker_Options is presented uninterpreted as
an argument to the link command, unless it contains Ascii.NUL characters.
NUL characters if they appear act as argument separators, so for example
pragma Linker_Options ("-labc" & ASCII.Nul & "-ldef");
causes two separate arguments -labc and -ldef to be passed to the
linker. The order of linker options is preserved for a given unit. The final
list of options passed to the linker is in reverse order of the elaboration
order. For example, linker options fo a body always appear before the options
from the corresponding package spec.
78. The contents of the visible part of package
Interfaces and its language-defined descendants. See B.2(1).
|
79. Implementation-defined children of package
Interfaces. The contents of the visible part of package
Interfaces. See B.2(11).
|
80. The types Floating, Long_Floating,
Binary, Long_Binary, Decimal_ Element, and
COBOL_Character; and the initialization of the variables
Ada_To_COBOL and COBOL_To_Ada, in
Interfaces.COBOL. See B.4(50).
|
FloatingLong_FloatingBinaryLong_BinaryDecimal_ElementCOBOL_CharacterFor initialization, see the file i-cobol.ads in the distributed library.
| 81. Support for access to machine instructions. See C.1(1). |
| 82. Implementation-defined aspects of access to machine operations. See C.1(9). |
| 83. Implementation-defined aspects of interrupts. See C.3(2). |
Ada.Interrupt_Names in source file a-intnam.ads for details
on the interrupts supported on a particular target.
| 84. Implementation-defined aspects of pre-elaboration. See C.4(13). |
85. The semantics of pragma Discard_Names. See C.5(7).
|
Discard_Names causes names of enumeration literals to
be suppressed. In the presence of this pragma, the Image attribute
provides the image of the Pos of the literal, and Value accepts
Pos values.
86. The result of the Task_Identification.Image
attribute. See C.7.1(7).
|
87. The value of Current_Task when in a protected entry
or interrupt handler. See C.7.1(17).
|
Current_Task is undefined.
88. The effect of calling Current_Task from an entry
body or interrupt handler. See C.7.1(19).
|
Current_Task from an entry body or
interrupt handler is to return the identification of the task currently
executing the code.
89. Implementation-defined aspects of
Task_Attributes. See C.7.2(19).
|
Task_Attributes.
90. Values of all Metrics. See D(2).
|
-gnatG can be
used to determine the exact sequence of operating systems calls made
to implement various tasking constructs. Together with appropriate
information on the performance of the underlying operating system,
on the exact target in use, this information can be used to determine
the required metrics.
91. The declarations of Any_Priority and
Priority. See D.1(11).
|
| 92. Implementation-defined execution resources. See D.1(15). |
| 93. Whether, on a multiprocessor, a task that is waiting for access to a protected object keeps its processor busy. See D.2.1(3). |
| 94. The affect of implementation defined execution resources on task dispatching. See D.2.1(9). |
95. Implementation-defined policy_identifiers allowed
in a pragma Task_Dispatching_Policy. See D.2.2(3).
|
| 96. Implementation-defined aspects of priority inversion. See D.2.2(16). |
| 97. Implementation defined task dispatching. See D.2.2(18). |
98. Implementation-defined policy_identifiers allowed
in a pragma Locking_Policy. See D.3(4).
|
Inheritance_Locking. On targets that support this policy, locking
is implemented by inheritance, i.e. the task owning the lock operates
at a priority equal to the highest priority of any task currently
requesting the lock.
| 99. Default ceiling priorities. See D.3(10). |
System.Interrupt_Priority'Last as described in the Ada 95
Reference Manual D.3(10),
| 100. The ceiling of any protected object used internally by the implementation. See D.3(16). |
System.Priority'Last.
| 101. Implementation-defined queuing policies. See D.4(1). |
| 102. On a multiprocessor, any conditions that cause the completion of an aborted construct to be delayed later than what is specified for a single processor. See D.6(3). |
| 103. Any operations that implicitly require heap storage allocation. See D.7(8). |
104. Implementation-defined aspects of pragma
Restrictions. See D.7(20).
|
105. Implementation-defined aspects of package
Real_Time. See D.8(17).
|
Real_Time.
106. Implementation-defined aspects of
delay_statements. See D.9(8).
|
| 107. The upper bound on the duration of interrupt blocking caused by the implementation. See D.12(5). |
| 108. The means for creating and executing distributed programs. See E(5). |
| 109. Any events that can result in a partition becoming inaccessible. See E.1(7). |
| 110. The scheduling policies, treatment of priorities, and management of shared resources between partitions in certain cases. See E.1(11). |
| 111. Events that cause the version of a compilation unit to change. See E.3(5). |
| 112. Whether the execution of the remote subprogram is immediately aborted as a result of cancellation. See E.4(13). |
| 113. Implementation-defined aspects of the PCS. See E.5(25). |
| 114. Implementation-defined interfaces in the PCS. See E.5(26). |
115. The values of named numbers in the package
Decimal. See F.2(7).
|
Max_ScaleMin_ScaleMin_DeltaMax_DeltaMax_Decimal_Digits
116. The value of Max_Picture_Length in the package
Text_IO.Editing. See F.3.3(16).
|
117. The value of Max_Picture_Length in the package
Wide_Text_IO.Editing. See F.3.4(5).
|
| 118. The accuracy actually achieved by the complex elementary functions and by other complex arithmetic operations. See G.1(1). |
119. The sign of a zero result (or a component thereof) from
any operator or function in Numerics.Generic_Complex_Types, when
Real'Signed_Zeros is True. See G.1.1(53).
|
120. The sign of a zero result (or a component thereof) from
any operator or function in
Numerics.Generic_Complex_Elementary_Functions, when
Real'Signed_Zeros is True. See G.1.2(45).
|
| 121. Whether the strict mode or the relaxed mode is the default. See G.2(2). |
| 122. The result interval in certain cases of fixed-to-float conversion. See G.2.1(10). |
123. The result of a floating point arithmetic operation in
overflow situations, when the Machine_Overflows attribute of the
result type is False. See G.2.1(13).
|
| 124. The result interval for division (or exponentiation by a negative exponent), when the floating point hardware implements division as multiplication by a reciprocal. See G.2.1(16). |
| 125. The definition of close result set, which determines the accuracy of certain fixed point multiplications and divisions. See G.2.3(5). |
126. Conditions on a universal_real operand of a fixed
point multiplication or division for which the result shall be in the
perfect result set. See G.2.3(22).
|
127. The result of a fixed point arithmetic operation in
overflow situations, when the Machine_Overflows attribute of the
result type is False. See G.2.3(27).
|
Machine_Overflows is True for fixed-point
types.
128. The result of an elementary function reference in
overflow situations, when the Machine_Overflows attribute of the
result type is False. See G.2.4(4).
|
| 129. The value of the angle threshold, within which certain elementary functions, complex arithmetic operations, and complex elementary functions yield results conforming to a maximum relative error bound. See G.2.4(10). |
| 130. The accuracy of certain elementary functions for parameters beyond the angle threshold. See G.2.4(10). |
131. The result of a complex arithmetic operation or complex
elementary function reference in overflow situations, when the
Machine_Overflows attribute of the corresponding real type is
False. See G.2.6(5).
|
| 132. The accuracy of certain complex arithmetic operations and certain complex elementary functions for parameters (or components thereof) beyond the angle threshold. See G.2.6(8). |
| 133. Information regarding bounded errors and erroneous execution. See H.2(1). |
134. Implementation-defined aspects of pragma
Inspection_Point. See H.3.2(8).
|
Inspection_Point ensures that the variable is live and can
be examined by the debugger at the inspection point.
135. Implementation-defined aspects of pragma
Restrictions. See H.4(25).
|
Restrictions. The
use of pragma Restrictions [No_Exceptions] has no effect on the
generated code. Checks must suppressed by use of pragma Suppress.
136. Any restrictions on pragma Restrictions. See
H.4(27).
|
Restrictions.
GNAT allows a user application program to write the declaration:
pragma Import (Intrinsic, name);
providing that the name corresponds to one of the implemented intrinsic subprograms in GNAT, and that the parameter profile of the referenced subprogram meets the requirements. This chapter describes the set of implemented intrinsic subprograms, and the requirements on parameter profiles. Note that no body is supplied; as with other uses of pragma Import, the body is supplied elsewhere (in this case by the compiler itself). Note that any use of this feature is potentially non-portable, since the Ada standard does not require Ada compilers to implement this feature.
All the predefined numeric operators in package Standard
in pragma Import (Intrinsic,..)
declarations. In the binary operator case, the operands must have the same
size. The operand or operands must also be appropriate for
the operator. For example, for addition, the operands must
both be floating-point or both be fixed-point, and the
right operand for "**" must have a root type of
Standard.Integer'Base.
You can use an intrinsic operator declaration as in the following example:
type Int1 is new Integer;
type Int2 is new Integer;
function "+" (X1 : Int1; X2 : Int2) return Int1;
function "+" (X1 : Int1; X2 : Int2) return Int2;
pragma Import (Intrinsic, "+");
This declaration would permit “mixed mode” arithmetic on items
of the differing types Int1 and Int2.
It is also possible to specify such operators for private types, if the
full views are appropriate arithmetic types.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Source_Info. The only useful use of the
intrinsic import in this case is the one in this unit, so an
application program should simply call the function
GNAT.Source_Info.Enclosing_Entity to obtain the name of
the current subprogram, package, task, entry, or protected subprogram.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Current_Exception. The only useful
use of the intrinsic import in this case is the one in this unit,
so an application program should simply call the function
GNAT.Current_Exception.Exception_Information to obtain
the exception information associated with the current exception.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Current_Exception. The only useful
use of the intrinsic import in this case is the one in this unit,
so an application program should simply call the function
GNAT.Current_Exception.Exception_Message to obtain
the message associated with the current exception.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Current_Exception. The only useful
use of the intrinsic import in this case is the one in this unit,
so an application program should simply call the function
GNAT.Current_Exception.Exception_Name to obtain
the name of the current exception.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Source_Info. The only useful use of the
intrinsic import in this case is the one in this unit, so an
application program should simply call the function
GNAT.Source_Info.File to obtain the name of the current
file.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Source_Info. The only useful use of the
intrinsic import in this case is the one in this unit, so an
application program should simply call the function
GNAT.Source_Info.Line to obtain the number of the current
source line.
In standard Ada 95, the Rotate_Left function is available only
for the predefined modular types in package Interfaces. However, in
GNAT it is possible to define a Rotate_Left function for a user
defined modular type or any signed integer type as in this example:
function Shift_Left
(Value : My_Modular_Type;
Amount : Natural)
return My_Modular_Type;
The requirements are that the profile be exactly as in the example
above. The only modifications allowed are in the formal parameter
names, and in the type of Value and the return type, which
must be the same, and must be either a signed integer type, or
a modular integer type with a binary modulus, and the size must
be 8. 16, 32 or 64 bits.
A Rotate_Right function can be defined for any user defined
binary modular integer type, or signed integer type, as described
above for Rotate_Left.
A Shift_Left function can be defined for any user defined
binary modular integer type, or signed integer type, as described
above for Rotate_Left.
A Shift_Right function can be defined for any user defined
binary modular integer type, or signed integer type, as described
above for Rotate_Left.
A Shift_Right_Arithmetic function can be defined for any user
defined binary modular integer type, or signed integer type, as described
above for Rotate_Left.
This intrinsic subprogram is used in the implementation of the
library routine GNAT.Source_Info. The only useful use of the
intrinsic import in this case is the one in this unit, so an
application program should simply call the function
GNAT.Source_Info.Source_Location to obtain the current
source file location.
This section describes the representation clauses accepted by GNAT, and their effect on the representation of corresponding data objects.
GNAT fully implements Annex C (Systems Programming). This means that all the implementation advice sections in chapter 13 are fully implemented. However, these sections only require a minimal level of support for representation clauses. GNAT provides much more extensive capabilities, and this section describes the additional capabilities provided.
GNAT requires that all alignment clauses specify a power of 2, and all default alignments are always a power of 2. The default alignment values are as follows:
Pack is
used and all fields are packable (see separate section on pragma Pack),
then the resulting alignment is 1.
A special case is when the size of the record is given explicitly, or a full record representation clause is given, and the size of the record is 2, 4, or 8 bytes. In this case, an alignment is chosen to match the size of the record. For example, if we have:
type Small is record
A, B : Character;
end record;
then the default alignment of the record type Small is 2, not 1. This
leads to more efficient code when the record is treated as a unit, and also
allows the type to specified as Atomic on architectures requiring
strict alignment.
An alignment clause may always specify a larger alignment than the default value, up to some maximum value dependent on the target (obtainable by using the attribute reference System'Maximum_Alignment). The only case in which it is permissible to specify a smaller alignment than the default value is in the case of a record for which a record representation clause is given. In this case, packable fields for which a component clause is given still result in a default alignment corresponding to the original type, but this may be overridden, since these components in fact only require an alignment of one byte. For example, given
type v is record
a : integer;
end record;
for v use record
a at 0 range 0 .. 31;
end record;
for v'alignment use 1;
The default alignment for the type v is 4, as a result of the
integer field in the record, but since this field is placed with a
component clause, it is permissible, as shown, to override the default
alignment of the record to a smaller value.
The default size of types is as specified in the reference manual. For objects, GNAT will generally increase the type size so that the object size is a multiple of storage units, and also a multiple of the alignment. For example
type Smallint is range 1 .. 6;
type Rec is record
y1 : integer;
y2 : boolean;
end record;
In this example, Smallint
has a size of 3, as specified by the RM rules,
but objects of this type will have a size of 8,
since objects by default occupy an integral number
of storage units. On some targets, notably older
versions of the Digital Alpha, the size of stand
alone objects of this type may be 32, reflecting
the inability of the hardware to do byte load/stores.
Similarly, the size of type Rec is 40 bits, but
the alignment is 4, so objects of this type will have
their size increased to 64 bits so that it is a multiple
of the alignment. The reason for this decision, which is
in accordance with the specific note in RM 13.3(43):
A Size clause should be supported for an object if the specified
Size is at least as large as its subtype's Size, and corresponds
to a size in storage elements that is a multiple of the object's
Alignment (if the Alignment is nonzero).
An explicit size clause may be used to override the default size by increasing it. For example, if we have:
type My_Boolean is new Boolean;
for My_Boolean'Size use 32;
then objects of this type will always be 32 bits long. In the case of discrete types, the size can be increased up to 64 bits, with the effect that the entire specified field is used to hold the value, sign- or zero-extended as appropriate. If more than 64 bits is specified, then padding space is allocated after the value, and a warning is issued that there are unused bits.
Similarly the size of records and arrays may be increased, and the effect is to add padding bits after the value. This also causes a warning message to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a Size in bits, this corresponds to an object of size 256 megabytes (minus one). This limitation is true on all targets. The reason for this limitation is that it improves the quality of the code in many cases if it is known that a Size value can be accommodated in an object of type Integer.
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified as required. Any tasks requiring different task stack sizes from the default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in connection with interfacing where no object will ever be allocated. If you compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the case of such an access declaration.
An issue arises in the case of variant record objects of whether Size gives information about a particular variant, or the maximum size required for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant requires 16 bits, and the False variant requires 8 bits. In the above example, both V1 and V2 contain the False variant, which is only 8 bits long. However, the result of running the program is:
8
16
The reason for the difference here is that the discriminant value of V1 is fixed, and will always be False. It is not possible to assign a True variant value to V1, therefore 8 bits is sufficient. On the other hand, in the case of V2, the initial discriminant value is False (from the default), but it is possible to assign a True variant value to V2, therefore 16 bits must be allocated for V2 in the general case, even fewer bits may be needed at any particular point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current variant, rather than the size of the largest variant. This can be achieved in GNAT by making use of the fact that in the case of a subprogram parameter, GNAT does indeed return the size of the current variant (because a subprogram has no way of knowing how much space is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
In the case of scalars with a range starting at other than zero, it is possible in some cases to specify a size smaller than the default minimum value, and in such cases, GNAT uses an unsigned biased representation, in which zero is used to represent the lower bound, and successive values represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This limitation is true on all targets. The reason for this
limitation is that it improves the quality of the code in many cases
if it is known that a Size value can be accommodated in an object of
type Integer.
Next: Size of Variant Record Objects,
Previous: Size Clauses,
Up: Representation Clauses and Pragmas
6.3 Storage_Size Clauses
For tasks, the Storage_Size clause specifies the amount of space
to be allocated for the task stack. This cannot be extended, and if the
stack is exhausted, then Storage_Error will be raised if stack
checking is enabled. If the default size of 20K bytes is insufficient,
then you need to use a Storage_Size attribute definition clause,
or a Storage_Size pragma in the task definition to set the
appropriate required size. A useful technique is to include in every
task definition a pragma of the form:
pragma Storage_Size (Default_Stack_Size);
Then Default_Stack_Size can be defined in a global package, and modified
as required. Any tasks requiring different task stack sizes from the
default can have an appropriate alternative reference in the pragma.
For access types, the Storage_Size clause specifies the maximum
space available for allocation of objects of the type. If this space is
exceeded then Storage_Error will be raised by an allocation attempt.
In the case where the access type is declared local to a subprogram, the
use of a Storage_Size clause triggers automatic use of a special
predefined storage pool (System.Pool_Size) that ensures that all
space for the pool is automatically reclaimed on exit from the scope in
which the type is declared.
A special case recognized by the compiler is the specification of a
Storage_Size of zero for an access type. This means that no
items can be allocated from the pool, and this is recognized at compile
time, and all the overhead normally associated with maintaining a fixed
size storage pool is eliminated. Consider the following example:
procedure p is
type R is array (Natural) of Character;
type P is access all R;
for P'Storage_Size use 0;
-- Above access type intended only for interfacing purposes
y : P;
procedure g (m : P);
pragma Import (C, g);
-- ...
begin
-- ...
y := new R;
end;
As indicated in this example, these dummy storage pools are often useful in
connection with interfacing where no object will ever be allocated. If you
compile the above example, you get the warning:
p.adb:16:09: warning: allocation from empty storage pool
p.adb:16:09: warning: Storage_Error will be raised at run time
Of course in practice, there will not be any explicit allocators in the
case of such an access declaration.
6.4 Size of Variant Record Objects
An issue arises in the case of variant record objects of whether Size gives
information about a particular variant, or the maximum size required
for any variant. Consider the following program
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V1 : R1 (False);
V2 : R1;
begin
Put_Line (Integer'Image (V1'Size));
Put_Line (Integer'Image (V2'Size));
end q;
Here we are dealing with a variant record, where the True variant
requires 16 bits, and the False variant requires 8 bits.
In the above example, both V1 and V2 contain the False variant,
which is only 8 bits long. However, the result of running the
program is:
8
16
The reason for the difference here is that the discriminant value of
V1 is fixed, and will always be False. It is not possible to assign
a True variant value to V1, therefore 8 bits is sufficient. On the
other hand, in the case of V2, the initial discriminant value is
False (from the default), but it is possible to assign a True
variant value to V2, therefore 16 bits must be allocated for V2
in the general case, even fewer bits may be needed at any particular
point during the program execution.
As can be seen from the output of this program, the 'Size
attribute applied to such an object in GNAT gives the actual allocated
size of the variable, which is the largest size of any of the variants.
The Ada Reference Manual is not completely clear on what choice should
be made here, but the GNAT behavior seems most consistent with the
language in the RM.
In some cases, it may be desirable to obtain the size of the current
variant, rather than the size of the largest variant. This can be
achieved in GNAT by making use of the fact that in the case of a
subprogram parameter, GNAT does indeed return the size of the current
variant (because a subprogram has no way of knowing how much space
is actually allocated for the actual).
Consider the following modified version of the above program:
with Text_IO; use Text_IO;
procedure q is
type R1 (A : Boolean := False) is record
case A is
when True => X : Character;
when False => null;
end case;
end record;
V2 : R1;
function Size (V : R1) return Integer is
begin
return V'Size;
end Size;
begin
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
V2 := (True, 'x');
Put_Line (Integer'Image (V2'Size));
Put_Line (Integer'IMage (Size (V2)));
end q;
The output from this program is
16
8
16
16
Here we see that while the 'Size attribute always returns
the maximum size, regardless of the current variant value, the
Size function does indeed return the size of the current
variant value.
Next: Value_Size and Object_Size Clauses,
Previous: Size of Variant Record Objects,
Up: Representation Clauses and Pragmas
6.5 Biased Representation
In the case of scalars with a range starting at other than zero, it is
possible in some cases to specify a size smaller than the default minimum
value, and in such cases, GNAT uses an unsigned biased representation,
in which zero is used to represent the lower bound, and successive values
represent successive values of the type.
For example, suppose we have the declaration:
type Small is range -7 .. -4;
for Small'Size use 2;
Although the default size of type Small is 4, the Size
clause is accepted by GNAT and results in the following representation
scheme:
-7 is represented as 2#00#
-6 is represented as 2#01#
-5 is represented as 2#10#
-4 is represented as 2#11#
Biased representation is only used if the specified Size clause
cannot be accepted in any other manner. These reduced sizes that force
biased representation can be used for all discrete types except for
enumeration types for which a representation clause is given.
Next: Component_Size Clauses,
Previous: Biased Representation,
Up: Representation Clauses and Pragmas
6.6 Value_Size and Object_Size Clauses
In Ada 95, the Size of a discrete type is the minimum number of bits
required to hold values of the type. Although this interpretation was
allowed in Ada 83, it was not required, and this requirement in practice
can cause some significant difficulties. For example, in most Ada 83
compilers, Natural'Size was 32. However, in Ada-95,
Natural'Size is
typically 31. This means that code may change in behavior when moving
from Ada 83 to Ada 95. For example, consider:
type Rec is record;
A : Natural;
B : Natural;
end record;
for Rec use record
for A use at 0 range 0 .. Natural'Size - 1;
for B use at 0 range Natural'Size .. 2 * Natural'Size - 1;
end record;
In the above code, since the typical size of Natural objects
is 32 bits and Natural'Size is 31, the above code can cause
unexpected inefficient packing in Ada 95, and in general there are
surprising cases where the fact that the object size can exceed the
size of the type causes surprises.
To help get around this problem GNAT provides two implementation
dependent attributes Value_Size and Object_Size. When
applied to a type, these attributes yield the size of the type
(corresponding to the RM defined size attribute), and the size of
objects of the type respectively.
The Object_Size is used for determining the default size of
objects and components. This size value can be referred to using the
Object_Size attribute. The phrase “is used” here means that it is
the basis of the determination of the size. The backend is free to
pad this up if necessary for efficiency, e.g. an 8-bit stand-alone
character might be stored in 32 bits on a machine with no efficient
byte access instructions such as the Alpha.
The default rules for the value of Object_Size for fixed-point and
discrete types are as follows:
Object_Size for base subtypes reflect the natural hardware
size in bits (run the utility gnatpsta to find those values for numeric types).
Enumeration types and fixed-point base subtypes have 8, 16, 32 or 64
bits for this size, depending on the range of values to be stored.
Object_Size of a subtype is the same as the
Object_Size of
the type from which it is obtained.
Object_Size of a derived base type is copied from the parent
base type, and the Object_Size of a derived first subtype is copied
from the parent first subtype.
The Value_Size attribute
is the number of bits required to store a value
of the type. This size can be referred to using the Value_Size
attribute. This value is used to determine how tightly to pack
records or arrays with components of this type, and also affects
the semantics of unchecked conversion (unchecked conversions where
the Value_Size values differ generate a warning, and are potentially
target dependent).
The default rules for the value of Value_Size are as follows:
Value_Size for a base subtype is the minimum number of bits
required to store all values of the type (including the sign bit
only if negative values are possible).
Value_Size as the first subtype. This is a
consequence of RM 13.1(14) (“if two subtypes statically match,
then their subtype-specific aspects are the same”.)
Value_Size corresponding to the minimum
number of bits required to store all values of the subtype. For
dynamic bounds, it is assumed that the value can range down or up
to the corresponding bound of the ancestor
The RM defined attribute Size corresponds to the
Value_Size attribute.
The Size attribute may be defined for a first-named subtype. This sets
the Value_Size of
the first-named subtype to the given value, and the
Object_Size of this first-named subtype to the given value padded up
to an appropriate boundary. It is a consequence of the default rules
above that this Object_Size will apply to all further subtypes. On the
other hand, Value_Size is affected only for the first subtype, any
dynamic subtypes obtained from it directly, and any statically matching
subtypes. The Value_Size of any other static subtypes is not affected.
Value_Size and
Object_Size may be explicitly set for any subtype using
an attribute definition clause. Note that the use of these attributes
can cause the RM 13.1(14) rule to be violated. If two access types
reference aliased objects whose subtypes have differing Object_Size
values as a result of explicit attribute definition clauses, then it
is erroneous to convert from one access subtype to the other.
At the implementation level, Esize stores the Object_SIze and the
RM_Size field stores the Value_Size (and hence the value of the
Size attribute,
which, as noted above, is equivalent to Value_Size).
To get a feel for the difference, consider the following examples (note that in each case the base is short_short_integer with a size of 8):
Object_Size Value_Size
type x1 is range 0 .. 5; 8 3
type x2 is range 0 .. 5;
class="noindent">then objects of this type will always be 32 bits long. In the case of
discrete types, the size can be increased up to 64 bits, with the effect
that the entire specified field is used to hold the value, sign- or
zero-extended as appropriate. If more than 64 bits is specified, then
padding space is allocated after the value, and a warning is issued that
there are unused bits.
Similarly the size of records and arrays may be increased, and the effect
is to add padding bits after the value. This also causes a warning message
to be generated.
The largest Size value permitted in GNAT is 2**32−1. Since this is a
Size in bits, this corresponds to an object of size 256 megabytes (minus
one). This l