abu@software-lab.de

Perfection is attained
not when there is nothing left to add
but when there is nothing left to take away
(Antoine de Saint-Exupéry)

The PicoLisp Reference

(c) Software Lab. Alexander Burger

This document describes the concepts, data types, and kernel functions of the PicoLisp system.

This is not a Lisp tutorial. For an introduction to Lisp, a traditional Lisp book like "Lisp" by Winston/Horn (Addison-Wesley 1981) is recommended. Note, however, that there are significant differences between PicoLisp and Maclisp (and even greater differences to Common Lisp).

Please take a look at the PicoLisp Tutorial for an explanation of some aspects of PicoLisp, and scan through the list of Frequently Asked Questions (FAQ).


Introduction

PicoLisp is the result of a language design study, trying to answer the question "What is a minimal but useful architecture for a virtual machine?". Because opinions differ about what is meant by "minimal" and "useful", there are many answers to that question, and people might consider other solutions more "minimal" or more "useful". But from a practical point of view, PicoLisp has proven to be a valuable answer to that question.

First of all, PicoLisp is a virtual machine architecture, and then a programming language. It was designed in a "bottom up" way, and "bottom up" is also the most natural way to understand and to use it: Form Follows Function.

PicoLisp has been used in several commercial and research programming projects since 1988. Its internal structures are simple enough, allowing an experienced programmer always to fully understand what's going on under the hood, and its language features, efficiency and extensibility make it suitable for almost any practical programming task.

In a nutshell, emphasis was put on four design objectives. The PicoLisp system should be

Simple
The internal data structure should be as simple as possible. Only one single data structure is used to build all higher level constructs.
Unlimited
There are no limits imposed upon the language due to limitations of the virtual machine architecture. That is, there is no upper bound in symbol name length, number digit counts, stack depth, or data structure and buffer sizes, except for the total memory size of the host machine.
Dynamic
Behavior should be as dynamic as possible ("run"-time vs. "compile"-time). All decisions are delayed until runtime where possible. This involves matters like memory management, dynamic symbol binding, and late method binding.
Practical
PicoLisp is not just a toy of theoretical value. It is in use since 1988 in actual application development, research and production.


The PicoLisp Machine

An important point in the PicoLisp philosophy is the knowledge about the architecture and data structures of the internal machinery. The high-level constructs of the programming language directly map to that machinery, making the whole system both understandable and predictable.

This is similar to assembly language programming, where the programmer has complete control over the machine.


The Cell

The PicoLisp virtual machine is both simpler and more powerful than most current (hardware) processors. At the lowest level, it is constructed from a single data structure called "cell":


         +-----+-----+
         | CAR | CDR |
         +-----+-----+

A cell is a pair of machine words, which traditionally are called CAR and CDR in the Lisp terminology. These words can represent either a numeric value (scalar) or the address of another cell (pointer). All higher level data structures are built out of cells.

The type information of higher level data is contained in the pointers to these data. Assuming the implementation on a byte-addressed physical machine, and a pointer size of typically 4 bytes, each cell has a size of 8 bytes. Therefore, the pointer to a cell must point to an 8-byte boundary, and its bit-representation will look like:


      xxxxxxxxxxxxxxxxxxxxxxxxxxxxx000

(the 'x' means "don't care"). For the individual data types, the pointer is adjusted to point to other parts of a cell, in effect setting some of the lower three bits to non-zero values. These bits are then used by the interpreter to determine the data type.

In any case, bit(0) - the least significant of these bits - is reserved as a mark bit for garbage collection.

Initially, all cells in the memory are unused (free), and linked together to form a "free list". To create higher level data types at runtime, cells are taken from that free list, and returned by the garbage collector when they are no longer needed. All memory management is done via that free list; there are no additional buffers, string spaces or special memory areas, with two exceptions:


Data Types

On the virtual machine level, PicoLisp supports

They are all built from the single cell data structure, and all runtime data cannot consist of any other types than these three.

The following diagram shows the complete data type hierarchy, consisting of the three base types and the symbol variations:


                    cell
                     |
            +--------+--------+
            |        |        |
         Number    Symbol    Pair
                     |
                     |
   +--------+--------+--------+
   |        |        |        |
  NIL   Internal Transient External


Numbers

A number can represent a signed integral value of arbitrary size. The CARs of one or more cells hold the number's "digits" (each in the machine's word size), to store the number's binary representation.


         Number
         |
         V
      +-----+-----+
      | DIG |  |  |
      +-----+--+--+
               |
               V
            +-----+-----+
            | DIG |  |  |
            +-----+--+--+
                     |
                     V
                    ...

The first cell holds the least significant digit. The least significant bit of that digit represents the sign.

The pointer to a number points into the middle of the CAR, with an offset of 2 from the cell's start address. Therefore, the bit pattern of a number will be:


      xxxxxxxxxxxxxxxxxxxxxxxxxxxxx010

Thus, a number is recognized by the interpreter when bit(1) is non-zero.


Symbols

A symbol is more complex than a number. Each symbol has a value, and optionally a name and an arbitrary number of properties. The CDR of a symbol cell is also called VAL, and the CAR points to the symbol's tail. As a minimum, a symbol consists of a single cell, and has no name or properties:


            Symbol
            |
            V
      +-----+-----+
      |  /  | VAL |
      +-----+-----+

That is, the symbol's tail is empty (points to NIL, as indicated by the '/' character).

The pointer to a symbol points to the CDR of the cell, with an offset of 4 from the cell's start address. Therefore, the bit pattern of a symbol will be:


      xxxxxxxxxxxxxxxxxxxxxxxxxxxxx100

Thus, a symbol is recognized by the interpreter when bit(2) is non-zero.

A property is a key-value pair, represented by a cons pair in the symbol's tail. This is called a "property list". The property list may be terminated by a number representing the symbol's name. In the following example, a symbol with the name "abc" has three properties: A KEY/VAL pair, a cell with only a KEY, and another KEY/VAL pair.


            Symbol
            |
            V
      +-----+-----+
      |  |  | VAL |
      +--+--+-----+
         | tail
         |
         V                                                      name
         +-----+-----+     +-----+-----+     +-----+-----+     +-----+-----+
         |  |  |  ---+---> | KEY |  ---+---> |  |  |  ---+---> |'cba'|  /  |
         +--+--+-----+     +-----+-----+     +--+--+-----+     +-----+-----+
            |                                   |
            V                                   V
            +-----+-----+                       +-----+-----+
            | VAL | KEY |                       | VAL | KEY |
            +-----+-----+                       +-----+-----+

Each property in a symbol's tail is either a symbol (like the single KEY above, then it represents the boolean value T), or a cons pair with the property key in its CDR and the property value in its CAR. In both cases, the key should be a symbol, because searches in the property list are performed using pointer comparisons.

The name of a symbol is stored as a number at the end of the tail. It contains the characters of the name in UTF-8 encoding, using between one and three 8-bit-bytes per character. The first byte of the first character is stored in the lowest 8 bits of the number.

All symbols have the above structure, but depending on scope and accessibility there are actually four types of symbols: NIL, internal, transient and external symbols.


NIL

NIL is a special symbol which exists exactly once in the whole system. It is used

For that, NIL has a special structure:


      NIL:  /
            |
            V
      +-----+-----+-----+-----+
      |  /  |  /  |  /  |  /  |
      +-----+--+--+-----+-----+

The reason for that structure is NIL's dual nature both as a symbol and as a list:

These requirements are fulfilled by the above structure.


Internal Symbols

Internal Symbols are all those "normal" symbols, as they are used for function definitions and variable names. They are "interned" into an index structure, so that it is possible to find an internal symbol by searching for its name.

There cannot be two different internal symbols with the same name.

Initially, a new internal symbol's VAL is NIL.


Transient Symbols

Transient symbols are only interned into a index structure for a certain time (e.g. while reading the current source file), and are released after that. That means, a transient symbol cannot be accessed then by its name, and there may be several transient symbols in the system having the same name.

Transient symbols are used

Initially, a new transient symbol's VAL is that symbol itself.

A transient symbol without a name can be created with the box or new functions.


External Symbols

External symbols reside in a database file (or a similar resources, see *Ext), and are loaded into memory - and written back to the file - dynamically as needed, and transparently to the programmer. They are kept in memory ("cached") as long as they are accessible ("referred to") from other parts of the program, or when they were modified but not yet written to the database file (by commit).

The interpreter recognizes external symbols internally by an additional tag bit in the tail structure.

There cannot be two different external symbols with the same name. External symbols are maintained in index structures while they are loaded into memory, and have their external location (disk file and block offset) directly coded into their names (more details here).

Initially, a new external symbol's VAL is NIL, unless otherwise specified at creation time.


Lists

A list is a sequence of one or more cells (cons pairs), holding numbers, symbols, or cons pairs.


      |
      V
      +-----+-----+
      | any |  |  |
      +-----+--+--+
               |
               V
               +-----+-----+
               | any |  |  |
               +-----+--+--+
                        |
                        V
                        ...

Lists are used in PicoLisp to emulate composite data structures like arrays, trees, stacks or queues.

In contrast to lists, numbers and symbols are collectively called "Atoms".

Typically, the CDR of each cell in a list points to the following cell, except for the last cell which points to NIL. If, however, the CDR of the last cell points to an atom, that cell is called a "dotted pair" (because of its I/O syntax with a dot '.' between the two values).


Memory Management

The PicoLisp interpreter has complete knowledge of all data in the system, due to the type information associated with every pointer. Therefore, an efficient garbage collector mechanism can easily be implemented. PicoLisp employs a simple but fast mark-and-sweep garbage collector.

As the collection process is very fast (in the order of milliseconds per megabyte), it was not necessary to develop more complicated, time-consuming and error-prone garbage collection algorithms (e.g. incremental collection). A compacting garbage collector is also not necessary, because the single cell data type cannot cause heap fragmentation.


Programming Environment

Lisp was chosen as the programming language, because of its clear and simple structure.

In some previous versions, a Forth-like syntax was also implemented on top of a similar virtual machine (Lifo). Though that language was more flexible and expressive, the traditional Lisp syntax proved easier to handle, and the virtual machine can be kept considerably simpler. PicoLisp inherits the major advantages of classical Lisp systems like

In the following, some concepts and peculiarities of the PicoLisp language and environment are described.


Installation

PicoLisp supports two installation strategies: Local and Global.

Normally, if you didn't build PicoLisp yourself but installed it with your operating system's package manager, you will have a global installation. This allows system-wide access to the executable and library/documentation files.

To get a local installation, you can directly download the PicoLisp tarball, and follow the instructions in the INSTALL file.

A local installation will not interfere in any way with the world outside its directory. There is no need to touch any system locations, and you don't have to be root to install it. Many different versions - or local modifications - of PicoLisp can co-exist on a single machine.

Note that you are still free to have local installations along with a global installation, and invoke them explicitly as desired.

Most examples in the following apply to a global installation.


Invocation

When PicoLisp is invoked from the command line, an arbitrary number of arguments may follow the command name.

By default, each argument is the name of a file to be executed by the interpreter. If, however, the argument's first character is a hyphen '-', then the rest of that argument is taken as a Lisp function call (without the surrounding parentheses), and a hyphen by itself as an argument stops evaluation of the rest of the command line (it may be processed later using the argv and opt functions). This whole mechanism corresponds to calling (load T).

A special case is if the last argument is a single '+'. This will switch on debug mode (the *Dbg global variable) and discard the '+'.

As a convention, PicoLisp source files have the extension ".l".

Note that the PicoLisp executable itself does not expect or accept any command line flags or options (except the '+', see above). They are reserved for application programs.

The simplest and shortest invocation of PicoLisp does nothing, and exits immediately by calling bye:


$ picolisp -bye
$

In interactive mode, the PicoLisp interpreter (see load) will also exit when Ctrl-D is entered:


$ picolisp
: $                     # Typed Ctrl-D

To start up the standard PicoLisp environment, several files should be loaded. The most commonly used things are in "lib.l" and in a bunch of other files, which are in turn loaded by "ext.l". Thus, a typical call would be:


$ picolisp lib.l ext.l

The recommended way, however, is to call the "pil" shell script, which includes "lib.l" and "ext.l". Given that your current project is loaded by some file "myProject.l" and your startup function is main, your invocation would look like:


$ pil myProject.l -main

For interactive development it is recommended to enable debugging mode, to get the vi-style command line editor, single-stepping, tracing and other debugging utilities.


$ pil myProject.l -main +

This is - in a local installation - equivalent to


$ ./dbg myProject.l -main

or


$ ./pil myProject.l -main +

In any case, the directory part of the first file name supplied (normally, the path to "lib.l" as called by 'pil' or 'dbg') is remembered internally as the PicoLisp Home Directory. This path is later automatically substituted for any leading "@" character in file name arguments to I/O functions (see path).


Input/Output

In Lisp, each internal data structure has a well-defined external representation in human-readable format. All kinds of data can be written to a file, and restored later to their original form by reading that file.

In normal operation, the PicoLisp interpreter continuously executes an infinite "read-eval-print loop". It reads one expression at a time, evaluates it, and prints the result to the console. Any input into the system, like data structures and function definitions, is done in a consistent way no matter whether it is entered at the console or read from a file.

Comments can be embedded in the input stream with the hash # character. Everything up to the end of that line will be ignored by the reader.


: (* 1 2 3)  # This is a comment
-> 6

A comment spanning several lines may be enclosed between #{ and }#.

Here is the I/O syntax for the individual PicoLisp data types (numbers, symbols and lists) and for read-macros:


Numbers

A number consists of an arbitrary number of digits ('0' through '9'), optionally preceded by a sign character ('+' or '-'). Legal number input is:


: 7
-> 7
: -12345678901245678901234567890
-> -12345678901245678901234567890

Fixpoint numbers can be input by embedding a decimal point '.', and setting the global variable *Scl appropriately:


: *Scl
-> 0

: 123.45
-> 123
: 456.78
-> 457

: (setq *Scl 3)
-> 3
: 123.45
-> 123450
: 456.78
-> 456780

Thus, fixpoint input simply scales the number to an integer value corresponding to the number of digits in *Scl.

Formatted output of scaled fixpoint values can be done with the format and round functions:


: (format 1234567890 2)
-> "12345678.90"
: (format 1234567890 2 "." ",")
-> "12,345,678.90"


Symbols

The reader is able to recognize the individual symbol types from their syntactic form. A symbol name should - of course - not look like a legal number (see above).

In general, symbol names are case-sensitive. car is not the same as CAR.


NIL

Besides for standard normal form, NIL is also recognized as (), [] or "".


: NIL
-> NIL
: ()
-> NIL
: ""
-> NIL

Output will always appear as NIL.


Internal Symbols

Internal symbol names can consist of any printable (non-whitespace) character, except for the following meta characters:


   "  '  (  )  ,  [  ]  `  ~ { }

It is possible, though, to include these special characters into symbol names by escaping them with a backslash '\'.

The dot '.' has a dual nature. It is a meta character when standing alone, denoting a dotted pair, but can otherwise be used in symbol names.

As a rule, anything not recognized by the reader as another data type will be returned as an internal symbol.


Transient Symbols

A transient symbol is anything surrounded by double quotes '"'. With that, it looks - and can be used - like a string constant in other languages. However, it is a real symbol, and may be assigned a value or a function definition, and properties.

Initially, a transient symbol's value is that symbol itself, so that it does not need to be quoted for evaluation:


: "This is a string"
-> "This is a string"

However, care must be taken when assigning a value to a transient symbol. This may cause unexpected behavior:


: (setq "This is a string" 12345)
-> 12345
: "This is a string"
-> 12345

The name of a transient symbol can contain any character except the null-byte. A double quote character can be escaped with a backslash '\', and a backslash itself has to be escaped with another backslash. Control characters can be written with a preceding hat '^' character.


: "We^Ird\\Str\"ing"
-> "We^Ird\\Str\"ing"
: (chop @)
-> ("W" "e" "^I" "r" "d" "\\" "S" "t" "r" "\"" "i" "n" "g")

The index for transient symbols is cleared automatically before and after loading a source file, or it can be reset explicitly with the ==== function. With that mechanism, it is possible to create symbols with a local access scope, not accessible from other parts of the program.

A special case of transient symbols are anonymous symbols. These are symbols without name (see box, box? or new). They print as a dollar sign ($) followed by a decimal digit string (actually their machine address).


External Symbols

External symbol names are surrounded by braces ('{' and '}'). The characters of the symbol's name itself identify the physical location of the external object. This is

In both cases, the database file (and possibly the hypen) are omitted for the first (default) file.


Lists

Lists are surrounded by parentheses ('(' and ')').

(A) is a list consisting of a single cell, with the symbol A in its CAR, and NIL in its CDR.

(A B C) is a list consisting of three cells, with the symbols A, B and C respectively in their CAR, and NIL in the last cell's CDR.

(A . B) is a "dotted pair", a list consisting of a single cell, with the symbol A in its CAR, and B in its CDR.

PicoLisp has built-in support for reading and printing simple circular lists. If the dot in a dotted-pair notation is immediately followed by a closing parenthesis, it indicates that the CDR of the last cell points back to the beginning of that list.


: (let L '(a b c) (conc L L))
-> (a b c .)
: (cdr '(a b c .))
-> (b c a .)
: (cddddr '(a b c .))
-> (b c a .)

A similar result can be achieved with the function circ. Such lists must be used with care, because many functions won't terminate or will crash when given such a list.


Read-Macros

Read-macrose used for function definitions and variable names. They are "interned" into an index structure, so that it is possible to find an internal symbol by searching for its name.

There cannot be two different internal symbols with the same name.

Initially, a new internal symbol's VAL is NIL.


Transient Symbols

Transient symbols are only interned into a index structure for a certain time (e.g. while reading the current source file), and are released after that. That means, a transient symbol cannot be accessed then by its name, and there may be several transient symbols in the system having the same name.

Transient symbols are used

Initially, a new transient symbol's VAL is that symbol itself.

A transient symbol without a name can be created with the box or new functions.


External Symbols

External symbols reside in a database file (or a similar resources, see *Ext), and are loaded into memory - and written back to the file - dynamically as needed, and transparently to the programmer. They are kept in memory ("cached") as long as they are accessible ("referred to") from other parts of the program, or when they were modified but not yet written to the database file (by commit).

The interpreter recognizes external symbols internally by an additional tag bit in the tail structure.

There cannot be two different external symbols with the same name. External symbols are maintained in index structures while they are loaded into memory, and have their external location (disk file and block offset) directly coded into their names (more details here).

Initially, a new external symbol's VAL is NIL, unless otherwise specified at creation time.


Lists

A list is a sequence of one or more cells (cons pairs), holding numbers, symbols, or cons pairs.


      |
      V
      +-----+-----+
      | any |  |  |
      +-----+--+--+
               |
               V
               +-----+-----+
               | any |  |  |
               +-----+--+--+
                        |
                        V
                        ...

Lists are used in PicoLisp to emulate composite data structures like arrays, trees, stacks or queues.

In contrast to lists, numbers and symbols are collectively called "Atoms".

Typically, the CDR of each cell in a list points to the following cell, except for the last cell which points to NIL. If, however, the CDR of the last cell points to an atom, that cell is called a "dotted pair" (because of its I/O syntax with a dot '.' between the two values).


Memory Management

The PicoLisp interpreter has complete knowledge of all data in the system, due to the type information associated with every pointer. Therefore, an efficient garbage collector mechanism can easily be implemented. PicoLisp employs a simple but fast mark-and-sweep garbage collector.

As the collection process is very fast (in the order of milliseconds per megabyte), it was not necessary to develop more complicated, time-consuming and error-prone garbage collection algorithms (e.g. incremental collection). A compacting garbage collector is also not necessary, because the single cell data type cannot cause heap fragmentation.


Programming Environment

Lisp was chosen as the programming language, because of its clear and simple structure.

In some previous versions, a Forth-like syntax was also implemented on top of a similar virtual machine (Lifo). Though that language was more flexible and expressive, the traditional Lisp syntax proved easier to handle, and the virtual machine can be kept considerably simpler. PicoLisp inherits the major advantages of classical Lisp systems like

In the following, some concepts and peculiarities of the PicoLisp language and environment are described.


Installation

PicoLisp supports two installation strategies: Local and Global.

Normally, if you didn't build PicoLisp yourself but installed it with your operating system's package manager, you will have a global installation. This allows system-wide access to the executable and library/documentation files.

To get a local installation, you can directly download the PicoLisp tarball, and follow the instructions in the INSTALL file.

A local installation will not interfere in any way with the world outside its directory. There is no need to touch any system locations, and you don't have to be root to install it. Many different versions - or local modifications - of PicoLisp can co-exist on a single machine.

Note that you are still free to have local installations along with a global installation, and invoke them explicitly as desired.

Most examples in the following apply to a global installation.


Invocation

When PicoLisp is invoked from the command line, an arbitrary number of arguments may follow the command name.

By default, each argument is the name of a file to be executed by the interpreter. If, however, the argument's first character is a hyphen '-', then the rest of that argument is taken as a Lisp function call (without the surrounding parentheses), and a hyphen by itself as an argument stops evaluation of the rest of the command line (it may be processed later using the argv and opt functions). This whole mechanism corresponds to calling (load T).

A special case is if the last argument is a single '+'. This will switch on debug mode (the *Dbg global variable) and discard the '+'.

As a convention, PicoLisp source files have the extension ".l".

Note that the PicoLisp executable itself does not expect or accept any command line flags or options (except the '+', see above). They are reserved for application programs.

The simplest and shortest invocation of PicoLisp does nothing, and exits immediately by calling bye:


$ picolisp -bye
$

In interactive mode, the PicoLisp interpreter (see load) will also exit when Ctrl-D is entered:


$ picolisp
: $                     # Typed Ctrl-D

To start up the standard PicoLisp environment, several files should be loaded. The most commonly used things are in "lib.l" and in a bunch of other files, which are in turn loaded by "ext.l". Thus, a typical call would be:


$ picolisp lib.l ext.l

The recommended way, however, is to call the "pil" shell script, which includes "lib.l" and "ext.l". Given that your current project is loaded by some file "myProject.l" and your startup function is main, your invocation would look like:


$ pil myProject.l -main

For interactive development it is recommended to enable debugging mode, to get the vi-style command line editor, single-stepping, tracing and other debugging utilities.


$ pil myProject.l -main +

This is - in a local installation - equivalent to


$ ./dbg myProject.l -main

or


$ ./pil myProject.l -main +

In any case, the directory part of the first file name supplied (normally, the path to "lib.l" as called by 'pil' or 'dbg') is remembered internally as the PicoLisp Home Directory. This path is later automatically substituted for any leading "@" character in file name arguments to I/O functions (see path).


Input/Output

In Lisp, each internal data structure has a well-defined external representation in human-readable format. All kinds of data can be written to a file, and restored later to their original form by reading that file.

In normal operation, the PicoLisp interpreter continuously executes an infinite "read-eval-print loop". It reads one expression at a time, evaluates it, and prints the result to the console. Any input into the system, like data structures and function definitions, is done in a consistent way no matter whether it is entered at the console or read from a file.

Comments can be embedded in the input stream with the hash # character. Everything up to the end of that line will be ignored by the reader.


: (* 1 2 3)  # This is a comment
-> 6

A comment spanning several lines may be enclosed between #{ and }#.

Here is the I/O syntax for the individual PicoLisp data types (numbers, symbols and lists) and for read-macros:


Numbers

A number consists of an arbitrary number of digits ('0' through '9'), optionally preceded by a sign character ('+' or '-'). Legal number input is:


: 7
-> 7
: -12345678901245678901234567890
-> -12345678901245678901234567890

Fixpoint numbers can be input by embedding a decimal point '.', and setting the global variable *Scl appropriately:


: *Scl
-> 0

: 123.45
-> 123
: 456.78
-> 457

: (setq *Scl 3)
-> 3
: 123.45
-> 123450
: 456.78
-> 456780

Thus, fixpoint input simply scales the number to an integer value corresponding to the number of digits in *Scl.

Formatted output of scaled fixpoint values can be done with the format and round functions:


: (format 1234567890 2)
-> "12345678.90"
: (format 1234567890 2 "." ",")
-> "12,345,678.90"


Symbols

The reader is able to recognize the individual symbol types from their syntactic form. A symbol name should - of course - not look like a legal number (see above).

In general, symbol names are case-sensitive. car is not the same as CAR.


NIL

Besides for standard normal form, NIL is also recognized as (), [] or "".


: NIL
-> NIL
: ()
-> NIL
: ""
-> NIL

Output will always appear as NIL.


Internal Symbols

Internal symbol names can consist of any printable (non-whitespace) character, except for the following meta characters:


   "  '  (  )  ,  [  ]  `  ~ { }

It is possible, though, to include these special characters into symbol names by escaping them with a backslash '\'.

The dot '.' has a dual nature. It is a meta character when standing alone, denoting a dotted pair, but can otherwise be used in symbol names.

As a rule, anything not recognized by the reader as another data type will be returned as an internal symbol.


Transient Symbols

A transient symbol is anything surrounded by double quotes '"'. With that, it looks - and can be used - like a string constant in other languages. However, it is a real symbol, and may be assigned a value or a function definition, and properties.

Initially, a transient symbol's value is that symbol itself, so that it does not need to be quoted for evaluation:


: "This is a string"
-> "This is a string"

However, care must be taken when assigning a value to a transient symbol. This may cause unexpected behavior:


: (setq "This is a string" 12345)
-> 12345
: "This is a string"
-> 12345

The name of a transient symbol can contain any character except the null-byte. A double quote character can be escaped with a backslash '\', and a backslash itself has to be escaped with another backslash. Control characters can be written with a preceding hat '^' character.


: "We^Ird\\Str\"ing"
-> "We^Ird\\Str\"ing"
: (chop @)
-> ("W" "e" "^I" "r" "d" "\\" "S" "t" "r" "\"" "i" "n" "g")

The index for transient symbols is cleared automatically before and after loading a source file, or it can be reset explicitly with the ==== function. With that mechanism, it is possible to create symbols with a local access scope, not accessible from other parts of the program.

A special case of transient symbols are anonymous symbols. These are symbols without name (see box, box? or new). They print as a dollar sign ($) followed by a decimal digit string (actually their machine address).


External Symbols

External symbol names are surrounded by braces ('{' and '}'). The characters of the symbol's name itself identify the physical location of the external object. This is

In both cases, the database file (and possibly the hypen) are omitted for the first (default) file.


Lists

Lists are surrounded by parentheses ('(' and ')').

(A) is a list consisting of a single cell, with the symbol A in its CAR, and NIL in its CDR.

(A B C) is a list consisting of three cells, with the symbols A, B and C respectively in their CAR, and NIL in the last cell's CDR.

(A . B) is a "dotted pair", a list consisting of a single cell, with the symbol A in its CAR, and B in its CDR.

PicoLisp has built-in support for reading and printing simple circular lists. If the dot in a dotted-pair notation is immediately followed by a closing parenthesis, it indicates that the CDR of the last cell points back to the beginning of that list.


: (let L '(a b c) (conc L L))
-> (a b c .)
: (cdr '(a b c .))
-> (b c a .)
: (cddddr '(a b c .))
-> (b c a .)

A similar result can be achieved with the function circ. Such lists must be used with care, because many functions won't terminate or will crash when given such a list.


Read-Macros

Read-macrose used for function definitions and variable names. They are "interned" into an index structure, so that it is possible to find an internal symbol by searching for its name.

There cannot be two different internal symbols with the same name.

Initially, a new internal symbol's VAL is NIL.


Transient Symbols

Transient symbols are only interned into a index structure for a certain time (e.g. while reading the current source file), and are released after that. That means, a transient symbol cannot be accessed then by its name, and there may be several transient symbols in the system having the same name.

Transient symbols are used

Initially, a new transient symbol's VAL is that symbol itself.

A transient symbol without a name can be created with the box or new functions.


External Symbols

External symbols reside in a database file (or a similar resources, see *Ext), and are loaded into memory - and written back to the file - dynamically as needed, and transparently to the programmer. They are kept in memory ("cached") as long as they are accessible ("referred to") from other parts of the program, or when they were modified but not yet written to the database file (by commit).

The interpreter recognizes external symbols internally by an additional tag bit in the tail structure.

There cannot be two different external symbols with the same name. External symbols are maintained in index structures while they are loaded into memory, and have their external location (disk file and block offset) directly coded into their names (more details here).

Initially, a new external symbol's VAL is NIL, unless otherwise specified at creation time.


Lists

A list is a sequence of one or more cells (cons pairs), holding numbers, symbols, or cons pairs.


      |
      V
      +-----+-----+
      | any |  |  |
      +-----+--+--+
               |
               V
               +-----+-----+
               | any |  |  |
               +-----+--+--+
                        |
                        V
                        ...

Lists are used in PicoLisp to emulate composite data structures like arrays, trees, stacks or queues.

In contrast to lists, numbers and symbols are collectively called "Atoms".

Typically, the CDR of each cell in a list points to the following cell, except for the last cell which points to NIL. If, however, the CDR of the last cell points to an atom, that cell is called a "dotted pair" (because of its I/O syntax with a dot '.' between the two values).


Memory Management

The PicoLisp interpreter has complete knowledge of all data in the system, due to the type information associated with every pointer. Therefore, an efficient garbage collector mechanism can easily be implemented. PicoLisp employs a simple but fast mark-and-sweep garbage collector.

As the collection process is very fast (in the order of milliseconds per megabyte), it was not necessary to develop more complicated, time-consuming and error-prone garbage collection algorithms (e.g. incremental collection). A compacting garbage collector is also not necessary, because the single cell data type cannot cause heap fragmentation.


Programming Environment

Lisp was chosen as the programming language, because of its clear and simple structure.

In some previous versions, a Forth-like syntax was also implemented on top of a similar virtual machine (Lifo). Though that language was more flexible and expressive, the traditional Lisp syntax proved easier to handle, and the virtual machine can be kept considerably simpler. PicoLisp inherits the major advantages of classical Lisp systems like

In the following, some concepts and peculiarities of the PicoLisp language and environment are described.


Installation

PicoLisp supports two installation strategies: Local and Global.

Normally, if you didn't build PicoLisp yourself but installed it with your operating system's package manager, you will have a global installation. This allows system-wide access to the executable and library/documentation files.

To get a local installation, you can directly download the PicoLisp tarball, and follow the instructions in the INSTALL file.

A local installation will not interfere in any way with the world outside its directory. There is no need to touch any system locations, and you don't have to be root to install it. Many different versions - or local modifications - of PicoLisp can co-exist on a single machine.

Note that you are still free to have local installations along with a global installation, and invoke them explicitly as desired.

Most examples in the following apply to a global installation.


Invocation

When PicoLisp is invoked from the command line, an arbitrary number of arguments may follow the command name.

By default, each argument is the name of a file to be executed by the interpreter. If, however, the argument's first character is a hyphen '-', then the rest of that argument is taken as a Lisp function call (without the surrounding parentheses), and a hyphen by itself as an argument stops evaluation of the rest of the command line (it may be processed later using the argv and opt functions). This whole mechanism corresponds to calling (load T).

A special case is if the last argument is a single '+'. This will switch on debug mode (the *Dbg global variable) and discard the '+'.

As a convention, PicoLisp source files have the extension ".l".

Note that the PicoLisp executable itself does not expect or accept any command line flags or options (except the '+', see above). They are reserved for application programs.

The simplest and shortest invocation of PicoLisp does nothing, and exits immediately by calling bye:


$ picolisp -bye
$

In interactive mode, the PicoLisp interpreter (see load) will also exit when Ctrl-D is entered:


$ picolisp
: $                     # Typed Ctrl-D

To start up the standard PicoLisp environment, several files should be loaded. The most commonly used things are in "lib.l" and in a bunch of other files, which are in turn loaded by "ext.l". Thus, a typical call would be:


$ picolisp lib.l ext.l

The recommended way, however, is to call the "pil" shell script, which includes "lib.l" and "ext.l". Given that your current project is loaded by some file "myProject.l" and your startup function is main, your invocation would look like:


$ pil myProject.l -main

For interactive development it is recommended to enable debugging mode, to get the vi-style command line editor, single-stepping, tracing and other debugging utilities.


$ pil myProject.l -main +

This is - in a local installation - equivalent to


$ ./dbg myProject.l -main

or


$ ./pil myProject.l -main +

In any case, the directory part of the first file name supplied (normally, the path to "lib.l" as called by 'pil' or 'dbg') is remembered internally as the PicoLisp Home Directory. This path is later automatically substituted for any leading "@" character in file name arguments to I/O functions (see path).


Input/Output

In Lisp, each internal data structure has a well-defined external representation in human-readable format. All kinds of data can be written to a file, and restored later to their original form by reading that file.

In normal operation, the PicoLisp interpreter continuously executes an infinite "read-eval-print loop". It reads one expression at a time, evaluates it, and prints the result to the console. Any input into the system, like data structures and function definitions, is done in a consistent way no matter whether it is entered at the console or read from a file.

Comments can be embedded in the input stream with the hash # character. Everything up to the end of that line will be ignored by the reader.


: (* 1 2 3)  # This is a comment
-> 6

A comment spanning several lines may be enclosed between #{ and }#.

Here is the I/O syntax for the individual PicoLisp data types (numbers, symbols and lists) and for read-macros:


Numbers

A number consists of an arbitrary number of digits ('0' through '9'), optionally preceded by a sign character ('+' or '-'). Legal number input is:


: 7
-> 7
: -12345678901245678901234567890
-> -12345678901245678901234567890

Fixpoint numbers can be input by embedding a decimal point '.', and setting the global variable *Scl appropriately:


: *Scl
-> 0

: 123.45
-> 123
: 456.78
-> 457

: (setq *Scl 3)
-> 3
: 123.45
-> 123450
: 456.78
-> 456780

Thus, fixpoint input simply scales the number to an integer value corresponding to the number of digits in *Scl.

Formatted output of scaled fixpoint values can be done with the format and round functions:


: (format 1234567890 2)
-> "12345678.90"
: (format 1234567890 2 "." ",")
-> "12,345,678.90"


Symbols

The reader is able to recognize the individual symbol types from their syntactic form. A symbol name should - of course - not look like a legal number (see above).

In general, symbol names are case-sensitive. car is not the same as CAR.


NIL

Besides for standard normal form, NIL is also recognized as (), [] or "".


: NIL
-> NIL
: ()
-> NIL
: ""
-> NIL

Output will always appear as NIL.


Internal Symbols

Internal symbol names can consist of any printable (non-whitespace) character, except for the following meta characters:


   "  '  (  )  ,  [  ]  `  ~ { }

It is possible, though, to include these special characters into symbol names by escaping them with a backslash '\'.

The dot '.' has a dual nature. It is a meta character when standing alone, denoting a dotted pair, but can otherwise be used in symbol names.

As a rule, anything not recognized by the reader as another data type will be returned as an internal symbol.


Transient Symbols

A transient symbol is anything surrounded by double quotes '"'. With that, it looks - and can be used - like a string constant in other languages. However, it is a real symbol, and may be assigned a value or a function definition, and properties.

Initially, a transient symbol's value is that symbol itself, so that it does not need to be quoted for evaluation:


: "This is a string"
-> "This is a string"

However, care must be taken when assigning a value to a transient symbol. This may cause unexpected behavior:


: (setq "This is a string" 12345)
-> 12345
: "This is a string"
-> 12345

The name of a transient symbol can contain any character except the null-byte. A double quote character can be escaped with a backslash '\', and a backslash itself has to be escaped with another backslash. Control characters can be written with a preceding hat '^' character.


: "We^Ird\\Str\"ing"
-> "We^Ird\\Str\"ing"
: (chop @)
-> ("W" "e" "^I" "r" "d" "\\" "S" "t" "r" "\"" "i" "n" "g")

The index for transient symbols is cleared automatically before and after loading a source file, or it can be reset explicitly with the ==== function. With that mechanism, it is possible to create symbols with a local access scope, not accessible from other parts of the program.

A special case of transient symbols are anonymous symbols. These are symbols without name (see box, box? or new). They print as a dollar sign ($) followed by a decimal digit string (actually their machine address).


External Symbols

External symbol names are surrounded by braces ('{' and '}'). The characters of the symbol's name itself identify the physical location of the external object. This is

In both cases, the database file (and possibly the hypen) are omitted for the first (default) file.


Lists

Lists are surrounded by parentheses ('(' and ')').

(A) is a list consisting of a single cell, with the symbol A in its CAR, and NIL in its CDR.

(A B C) is a list consisting of three cells, with the symbols A, B and C respectively in their CAR, and NIL in the last cell's CDR.

(A . B) is a "dotted pair", a list consisting of a single cell, with the symbol A in its CAR, and B in its CDR.

PicoLisp has built-in support for reading and printing simple circular lists. If the dot in a dotted-pair notation is immediately followed by a closing parenthesis, it indicates that the CDR of the last cell points back to the beginning of that list.


: (let L '(a b c) (conc L L))
-> (a b c .)
: (cdr '(a b c .))
-> (b c a .)
: (cddddr '(a b c .))
-> (b c a .)

A similar result can be achieved with the function circ. Such lists must be used with care, because many functions won't terminate or will crash when given such a list.


Read-Macros

Read-macrose used for function definitions and variable names. They are "interned" into an index structure, so that it is possible to find an internal symbol by searching for its name.

There cannot be two different internal symbols with the same name.

Initially, a new internal symbol's VAL is NIL.


Transient Symbols

Transient symbols are only interned into a index structure for a certain time (e.g. while reading the current source file), and are released after that. That means, a transient symbol cannot be accessed then by its name, and there may be several transient symbols in the system having the same name.

Transient symbols are used

Initially, a new transient symbol's VAL is that symbol itself.

A transient symbol without a name can be created with the box or new functions.


External Symbols

External symbols reside in a database file (or a similar resources, see *Ext), and are loaded into memory - and written back to the file - dynamically as needed, and transparently to the programmer. They are kept in memory ("cached") as long as they are accessible ("referred to") from other parts of the program, or when they were modified but not yet written to the database file (by commit).

The interpreter recognizes external symbols internally by an additional tag bit in the tail structure.

There cannot be two different external symbols with the same name. External symbols are maintained in index structures while they are loaded into memory, and have their external location (disk file and block offset) directly coded into their names (more details here).

Initially, a new external symbol's VAL is NIL, unless otherwise specified at creation time.


Lists

A list is a sequence of one or more cells (cons pairs), holding numbers, symbols, or cons pairs.


      |
      V
      +-----+-----+
      | any |  |  |
      +-----+--+--+
               |
               V
               +-----+-----+
               | any |  |  |
               +-----+--+--+
                        |
                        V
                        ...

Lists are used in PicoLisp to emulate composite data structures like arrays, trees, stacks or queues.

In contrast to lists, numbers and symbols are collectively called "Atoms".

Typically, the CDR of each cell in a list points to the following cell, except for the last cell which points to NIL. If, however, the CDR of the last cell points to an atom, that cell is called a "dotted pair" (because of its I/O syntax with a dot '.' between the two values).


Memory Management

The PicoLisp interpreter has complete knowledge of all data in the system, due to the type information associated with every pointer. Therefore, an efficient garbage collector mechanism can easily be implemented. PicoLisp employs a simple but fast mark-and-sweep garbage collector.

As the collection process is very fast (in the order of milliseconds per megabyte), it was not necessary to develop more complicated, time-consuming and error-prone garbage collection algorithms (e.g. incremental collection). A compacting garbage collector is also not necessary, because the single cell data type cannot cause heap fragmentation.


Programming Environment

Lisp was chosen as the programming language, because of its clear and simple structure.

In some previous versions, a Forth-like syntax was also implemented on top of a similar virtual machine (Lifo). Though that language was more flexible and expressive, the traditional Lisp syntax proved easier to handle, and the virtual machine can be kept considerably simpler. PicoLisp inherits the major advantages of classical Lisp systems like

In the following, some concepts and peculiarities of the PicoLisp language and environment are described.


Installation

PicoLisp supports two installation strategies: Local and Global.

Normally, if you didn't build PicoLisp yourself but installed it with your operating system's package manager, you will have a global installation. This allows system-wide access to the executable and library/documentation files.

To get a local installation, you can directly download the PicoLisp tarball, and follow the instructions in the INSTALL file.

A local installation will not interfere in any way with the world outside its directory. There is no need to touch any system locations, and you don't have to be root to install it. Many different versions - or local modifications - of PicoLisp can co-exist on a single machine.

Note that you are still free to have local installations along with a global installation, and invoke them explicitly as desired.

Most examples in the following apply to a global installation.


Invocation

When PicoLisp is invoked from the command line, an arbitrary number of arguments may follow the command name.

By default, each argument is the name of a file to be executed by the interpreter. If, however, the argument's first character is a hyphen '-', then the rest of that argument is taken as a Lisp function call (without the surrounding parentheses), and a hyphen by itself as an argument stops evaluation of the rest of the command line (it may be processed later using the argv and opt functions). This whole mechanism corresponds to calling (load T).

A special case is if the last argument is a single '+'. This will switch on debug mode (the *Dbg global variable) and discard the '+'.

As a convention, PicoLisp source files have the extension ".l".

Note that the PicoLisp executable itself does not expect or accept any command line flags or options (except the '+', see above). They are reserved for application programs.

The simplest and shortest invocation of PicoLisp does nothing, and exits immediately by calling bye:


$ picolisp -bye
$

In interactive mode, the PicoLisp interpreter (see load) will also exit when Ctrl-D is entered:


$ picolisp
: $                     # Typed Ctrl-D

To start up the standard PicoLisp environment, several files should be loaded. The most commonly used things are in "lib.l" and in a bunch of other files, which are in turn loaded by "ext.l". Thus, a typical call would be:


$ picolisp lib.l ext.l

The recommended way, however, is to call the "pil" shell script, which includes "lib.l" and "ext.l". Given that your current project is loaded by some file "myProject.l" and your startup function is main, your invocation would look like:


$ pil myProject.l -main

For interactive development it is recommended to enable debugging mode, to get the vi-style command line editor, single-stepping, tracing and other debugging utilities.


$ pil myProject.l -main +

This is - in a local installation - equivalent to


$ ./dbg myProject.l -main

or


$ ./pil myProject.l -main +

In any case, the directory part of the first file name supplied (normally, the path to "lib.l" as called by 'pil' or 'dbg') is remembered internally as the PicoLisp Home Directory. This path is later automatically substituted for any leading "@" character in file name arguments to I/O functions (see path).


Input/Output

In Lisp, each internal data structure has a well-defined external representation in human-readable format. All kinds of data can be written to a file, and restored later to their original form by reading that file.

In normal operation, the PicoLisp interpreter continuously executes an infinite "read-eval-print loop". It reads one expression at a time, evaluates it, and prints the result to the console. Any input into the system, like data structures and function definitions, is done in a consistent way no matter whether it is entered at the console or read from a file.

Comments can be embedded in the input stream with the hash # character. Everything up to the end of that line will be ignored by the reader.


: (* 1 2 3)  # This is a comment
-> 6

A comment spanning several lines may be enclosed between #{ and }#.

Here is the I/O syntax for the individual PicoLisp data types (numbers, symbols and lists) and for read-macros:


Numbers

A number consists of an arbitrary number of digits ('0' through '9'), optionally preceded by a sign character ('+' or '-'). Legal number input is:


: 7
-> 7
: -12345678901245678901234567890
-> -12345678901245678901234567890

Fixpoint numbers can be input by embedding a decimal point '.', and setting the global variable *Scl appropriately:


: *Scl
-> 0

: 123.45
-> 123
: 456.78
-> 457

: (setq *Scl 3)
-> 3
: 123.45
-> 123450
: 456.78
-> 456780

Thus, fixpoint input simply scales the number to an integer value corresponding to the number of digits in *Scl.

Formatted output of scaled fixpoint values can be done with the format and round functions:


: (format 1234567890 2)
-> "12345678.90"
: (format 1234567890 2 "." ",")
-> "12,345,678.90"


Symbols

The reader is able to recognize the individual symbol types from their syntactic form. A symbol name should - of course - not look like a legal number (see above).

In general, symbol names are case-sensitive. car is not the same as CAR.


NIL

Besides for standard normal form, NIL is also recognized as (), [] or "".


: NIL
-> NIL
: ()
-> NIL
: ""
-> NIL

Output will always appear as NIL.


Internal Symbols

Internal symbol names can consist of any printable (non-whitespace) character, except for the following meta characters:


   "  '  (  )  ,  [  ]  `  ~ { }

It is possible, though, to include these special characters into symbol names by escaping them with a backslash '\'.

The dot '.' has a dual nature. It is a meta character when standing alone, denoting a dotted pair, but can otherwise be used in symbol names.

As a rule, anything not recognized by the reader as another data type will be returned as an internal symbol.


Transient Symbols

A transient symbol is anything surrounded by double quotes '"'. With that, it looks - and can be used - like a string constant in other languages. However, it is a real symbol, and may be assigned a value or a function definition, and properties.

Initially, a transient symbol's value is that symbol itself, so that it does not need to be quoted for evaluation:


: "This is a string"
-> "This is a string"

However, care must be taken when assigning a value to a transient symbol. This may cause unexpected behavior:


: (setq "This is a string" 12345)
-> 12345
: "This is a string"
-> 12345

The name of a transient symbol can contain any character except the null-byte. A double quote character can be escaped with a backslash '\', and a backslash itself has to be escaped with another backslash. Control characters can be written with a preceding hat '^' character.


: "We^Ird\\Str\"ing"
-> "We^Ird\\Str\"ing"
: (chop @)
-> ("W" "e" "^I" "r" "d" "\\" "S" "t" "r" "\"" "i" "n" "g")

The index for transient symbols is cleared automatically before and after loading a source file, or it can be reset explicitly with the ==== function. With that mechanism, it is possible to create symbols with a local access scope, not accessible from other parts of the program.

A special case of transient symbols are anonymous symbols. These are symbols without name (see box, box? or new). They print as a dollar sign ($) followed by a decimal digit string (actually their machine address).


External Symbols

External symbol names are surrounded by braces ('{' and '}'). The characters of the symbol's name itself identify the physical location of the external object. This is

In both cases, the database file (and possibly the hypen) are omitted for the first (default) file.


Lists

Lists are surrounded by parentheses ('(' and ')').

(A) is a list consisting of a single cell, with the symbol A in its CAR, and NIL in its CDR.

(A B C) is a list consisting of three cells, with the symbols A, B and C respectively in their CAR, and NIL in the last cell's CDR.

(A . B) is a "dotted pair", a list consisting of a single cell, with the symbol A in its CAR, and B in its CDR.

PicoLisp has built-in support for reading and printing simple circular lists. If the dot in a dotted-pair notation is immediately followed by a closing parenthesis, it indicates that the CDR of the last cell points back to the beginning of that list.


: (let L '(a b c) (conc L L))
-> (a b c .)
: (cdr '(a b c .))
-> (b c a .)
: (cddddr '(a b c .))
-> (b c a .)

A similar result can be achieved with the function circ. Such lists must be used with care, because many functions won't terminate or will crash when given such a list.


Read-Macros

Read-macrose used for function definitions and variable names. They are "interned" into an index structure, so that it is possible to find an internal symbol by searching for its name.

There cannot be two different internal symbols with the same name.

Initially, a new internal symbol's VAL is NIL.


Transient Symbols

Transient symbols are only interned into a index structure for a certain time (e.g. while reading the current source file), and are released after that. That means, a transient symbol cannot be accessed then by its name, and there may be several transient symbols in the system having the same name.

Transient symbols are used

Initially, a new transient symbol's VAL is that symbol itself.

A transient symbol without a name can be created with the box or new functions.


External Symbols

External symbols reside in a database file (or a similar resources, see *Ext), and are loaded into memory - and written back to the file - dynamically as needed, and transparently to the programmer. They are kept in memory ("cached") as long as they are accessible ("referred to") from other parts of the program, or when they were modified but not yet written to the database file (by commit).

The interpreter recognizes external symbols internally by an additional tag bit in the tail structure.

There cannot be two different external symbols with the same name. External symbols are maintained in index structures while they are loaded into memory, and have their external location (disk file and block offset) directly coded into their names (more details here).

Initially, a new external symbol's VAL is NIL, unless otherwise specified at creation time.


Lists

A list is a sequence of one or more cells (cons pairs), holding numbers, symbols, or cons pairs.


      |
      V
      +-----+-----+
      | any |  |  |
      +-----+--+--+
               |
               V
               +-----+-----+
               | any |  |  |
               +-----+--+--+
                        |
                        V
                        ...

Lists are used in PicoLisp to emulate composite data structures like arrays, trees, stacks or queues.

In contrast to lists, numbers and symbols are collectively called "Atoms".

Typically, the CDR of each cell in a list points to the following cell, except for the last cell which points to NIL. If, however, the CDR of the last cell points to an atom, that cell is called a "dotted pair" (because of its I/O syntax with a dot '.' between the two values).


Memory Management

The PicoLisp interpreter has complete knowledge of all data in the system, due to the type information associated with every pointer. Therefore, an efficient garbage collector mechanism can easily be implemented. PicoLisp employs a simple but fast mark-and-sweep garbage collector.

As the collection process is very fast (in the order of milliseconds per megabyte), it was not necessary to develop more complicated, time-consuming and error-prone garbage collection algorithms (e.g. incremental collection). A compacting garbage collector is also not necessary, because the single cell data type cannot cause heap fragmentation.


Programming Environment

Lisp was chosen as the programming language, because of its clear and simple structure.

In some previous versions, a Forth-like syntax was also implemented on top of a similar virtual machine (Lifo). Though that language was more flexible and expressive, the traditional Lisp syntax proved easier to handle, and the virtual machine can be kept considerably simpler. PicoLisp inherits the major advantages of classical Lisp systems like

In the following, some concepts and peculiarities of the PicoLisp language and environment are described.


Installation

PicoLisp supports two installation strategies: Local and Global.

Normally, if you didn't build PicoLisp yourself but installed it with your operating system's package manager, you will have a global installation. This allows system-wide access to the executable and library/documentation files.

To get a local installation, you can directly download the PicoLisp tarball, and follow the instructions in the INSTALL file.

A local installation will not interfere in any way with the world outside its directory. There is no need to touch any system locations, and you don't have to be root to install it. Many different versions - or local modifications - of PicoLisp can co-exist on a single machine.

Note that you are still free to have local installations along with a global installation, and invoke them explicitly as desired.

Most examples in the following apply to a global installation.


Invocation

When PicoLisp is invoked from the command line, an arbitrary number of arguments may follow the command name.

By default, each argument is the name of a file to be executed by the interpreter. If, however, the argument's first character is a hyphen '-', then the rest of that argument is taken as a Lisp function call (without the surrounding parentheses), and a hyphen by itself as an argument stops evaluation of the rest of the command line (it may be processed later using the argv and opt functions). This whole mechanism corresponds to calling (load T).

A special case is if the last argument is a single '+'. This will switch on debug mode (the *Dbg global variable) and discard the '+'.

As a convention, PicoLisp source files have the extension ".l".

Note that the PicoLisp executable itself does not expect or accept any command line flags or options (except the '+', see above). They are reserved for application programs.

The simplest and shortest invocation of PicoLisp does nothing, and exits immediately by calling bye:


$ picolisp -bye
$

In interactive mode, the PicoLisp interpreter (see load) will also exit when Ctrl-D is entered:


$ picolisp
: $                     # Typed Ctrl-D

To start up the standard PicoLisp environment, several files should be loaded. The most commonly used things are in "lib.l" and in a bunch of other files, which are in turn loaded by "ext.l". Thus, a typical call would be:


$ picolisp lib.l ext.l

The recommended way, however, is to call the "pil" shell script, which includes "lib.l" and "ext.l". Given that your current project is loaded by some file "myProject.l" and your startup function is main, your invocation would look like:


$ pil myProject.l -main

For interactive development it is recommended to enable debugging mode, to get the vi-style command line editor, single-stepping, tracing and other debugging utilities.


$ pil myProject.l -main +

This is - in a local installation - equivalent to


$ ./dbg myProject.l -main

or


$ ./pil myProject.l -main +

In any case, the directory part of the first file name supplied (normally, the path to "lib.l" as called by 'pil' or 'dbg') is remembered internally as the PicoLisp Home Directory. This path is later automatically substituted for any leading "@" character in file name arguments to I/O functions (see path).


Input/Output

In Lisp, each internal data structure has a well-defined external representation in human-readable format. All kinds of data can be written to a file, and restored later to their original form by reading that file.

In normal operation, the PicoLisp interpreter continuously executes an infinite "read-eval-print loop". It reads one expression at a time, evaluates it, and prints the result to the console. Any input into the system, like data structures and function definitions, is done in a consistent way no matter whether it is entered at the console or read from a file.

Comments can be embedded in the input stream with the hash # character. Everything up to the end of that line will be ignored by the reader.


: (* 1 2 3)  # This is a comment
-> 6

A comment spanning several lines may be enclosed between #{ and }#.

Here is the I/O syntax for the individual PicoLisp data types (numbers, symbols and lists) and for read-macros:


Numbers

A number consists of an arbitrary number of digits ('0' through '9'), optionally preceded by a sign character ('+' or '-'). Legal number input is:


: 7
-> 7
: -12345678901245678901234567890
-> -12345678901245678901234567890

Fixpoint numbers can be input by embedding a decimal point '.', and setting the global variable *Scl appropriately:


: *Scl
-> 0

: 123.45
-> 123
: 456.78
-> 457

: (setq *Scl 3)
-> 3
: 123.45
-> 123450
: 456.78
-> 456780

Thus, fixpoint input simply scales the number to an integer value corresponding to the number of digits in *Scl.

Formatted output of scaled fixpoint values can be done with the format and round functions:


: (format 1234567890 2)
-> "12345678.90"
: (format 1234567890 2 "." ",")
-> "12,345,678.90"


Symbols

The reader is able to recognize the individual symbol types from their syntactic form. A symbol name should - of course - not look like a legal number (see above).

In general, symbol names are case-sensitive. car is not the same as CAR.


NIL

Besides for standard normal form, NIL is also recognized as (), [] or "".


: NIL
-> NIL
: ()
-> NIL
: ""
-> NIL

Output will always appear as NIL.


Internal Symbols

Internal symbol names can consist of any printable (non-whitespace) character, except for the following meta characters:


   "  '  (  )  ,  [  ]  `  ~ { }

It is possible, though, to include these special characters into symbol names by escaping them with a backslash '\'.

The dot '.' has a dual nature. It is a meta character when standing alone, denoting a dotted pair, but can otherwise be used in symbol names.

As a rule, anything not recognized by the reader as another data type will be returned as an internal symbol.


Transient Symbols

A transient symbol is anything surrounded by double quotes '"'. With that, it looks - and can be used - like a string constant in other languages. However, it is a real symbol, and may be assigned a value or a function definition, and properties.

Initially, a transient symbol's value is that symbol itself, so that it does not need to be quoted for evaluation:


: "This is a string"
-> "This is a string"

However, care must be taken when assigning a value to a transient symbol. This may cause unexpected behavior:


: (setq "This is a string" 12345)
-> 12345
: "This is a string"
-> 12345

The name of a transient symbol can contain any character except the null-byte. A double quote character can be escaped with a backslash '\', and a backslash itself has to be escaped with another backslash. Control characters can be written with a preceding hat '^' character.


: "We^Ird\\Str\"ing"
-> "We^Ird\\Str\"ing"
: (chop @)
-> ("W" "e" "^I" "r" "d" "\\" "S" "t" "r" "\"" "i" "n" "g")

The index for transient symbols is cleared automatically before and after loading a source file, or it can be reset explicitly with the ==== function. With that mechanism, it is possible to create symbols with a local access scope, not accessible from other parts of the program.

A special case of transient symbols are anonymous symbols. These are symbols without name (see box, box? or new). They print as a dollar sign ($) followed by a decimal digit string (actually their machine address).


External Symbols

External symbol names are surrounded by braces ('{' and '}'). The characters of the symbol's name itself identify the physical location of the external object. This is