The following are the changes from calc version 2.12.1 to date:

    Fixed minor typos in the 'version 2.12.0 to 2.12.0.8' section below.
    Made minor formatting changes as well.

    Changed use of ${Q} in the Makefile to avoid an make "feature"
    related to OpenBSD.  Added ${RM} make variable for make tools that
    do not have builtin defined terms.

    Removed the ECHO_PROG Makefile variable.  Also removed it from
    the sysinfo() custom function.

    Improved the support for cross-compiled environmens by using
    make symbols for all non-shell commands executed by Makefiles.

    Fixed a problem with the make chk awk script which failed under
    OS X 10.4.7.

    Fixed a few minor variables that were not set to default values in
    lower level Makefiles.

    Fixed a reference to a non-existent make variable in HOWTO.INSTALL.


The following are the changes from calc version 2.12.0 to 2.12.0.8:

    Fixed ellip.cal to deal with a calc syntax change that happened
    many ages ago but was never applied to this file until now.
    This bug was fixed by Ernest Bowen <ebowen at une dot edu dot au>.

    Fixed a problem where comments using # followed by a !, newline or
    another # works.  This bug was fixed by Ernest Bowen <ebowen at une
    dot edu dot au>.

    The show builtins display for functions with long descriptions
    is now broken into multi-line descriptions.

    The str functions, such as strcpy(s1, s2), will now copy as many
    characters as possible from s2 to s1, treating '\0' like any other
    character until the end of s2 is reached. If s2 is shorter than s1,
    a '\0' is inserted.

    The strcmp(s1, s2) builtin, for strings s1, s2: strcmp(s1, s2) == 0 now
    means the same as s1 == s2.

    The str(s) builtin has been changed so that it will return only the
    string formed by the characters of 's' up to the first '\0'.

    The substr(s, start, num) builtin has been changed so that '\0' characters
    are treated like any other.

    Fixed a bug where strcpy("", "a") used to cause a segmentation fault.
    This bug was fixed by Ernest Bowen <ebowen at une dot edu dot au>.

    Make minor change to natnumset.cal in how the tail variable is initialized.

    Fixed bugs in the strcmp, strncmp, strcpy, and strncpy help files.
    This bug was fixed by Ernest Bowen <ebowen at une dot edu dot au>.

    Added cal/screen.cal which Defines ANSI control sequences providing
    (i.e., cursor movement, changing foreground or background color,
    etc.) for VT100 terminals and terminal window emulators (i.e., xterm,
    Apple OS/X Terminal, etc.) that support them.  For example:

	; read screen
	; print green:"This is green. ":red:"This is red.":black

    Fixed a bug where too many open files returned E_FOPEN3.  Now
    a new error symbol F_MANYOPEN is used for too many open files.

    Added the builtin function fpathopen() to open a file while
    searching along a path:

    	; fd2 = fpathopen("tmp/date", "r", ".:~:~sc:/tmp:/var/tmp:/var")
	; print fd2
	"/var/tmp/date"

    By default, fpathopen() searches along CALCPATH.

    Added the calcpath() builtin function to return the current value
    of CALCPATH.

    Fixed prompt characters in the EXAMPLE section of help files.

    Fixed problems related to the protect function and its documentation.
    This bug was reported by David Gilham <davidgilham at gmail dot com>.
    This bug was fixed by Ernest Bowen <ebowen at une dot edu dot au>.

    Raised the limit of exponent in exponential notation.  It was set to
    arbitrary 1000000 (making 1e1000001 in invalid exponential notation
    value).  The exponent for exponential notation is now int(MAXLONG/10).
    On 32 bit machines, this means a limit of 214748364.  On 64 bit
    machines, this means 922337203685477580.  Of course, you may not
    have enough memory to hold such huge values, but if you did you can
    now express such values in exponential notation.

    Added log() builtin for base 10 logarithm.

    Fixed problems where internal use of libc strcpy() might have caused
    a buffer overflow.  Calc now only uses libc strcpy() when the source
    string is a constant.

    The calc STRING and STRINGHEAD now use the standard size_t (an unsigned
    type) length.  Calc mostly uses size_t in dealing with string lengths
    and object sizes when possible.

    Added ${CCWERR} make variable to allow one to force compiler warnings
    to be treated as errors.  The ${CC} make variable now uses ${CCWERR}
    however the ${LCC} (used by the Makefile test code for building hsrc
    files) does not use ${CCWERR}.  By default, ${CCWERR} is empty.
    In development Makefiles, we set CCWERR= -Werror to force us to
    address compiler warnings before the next release.

    The calc make variable, CALCPAGER, now defaults to CALCPAGER= less
    because the less utility is now very common.  Set CALCPAGER= more
    if you do not have less.

    Calc source had two styles of switch indentation.  Converted the
    style where case statements were indented with respect to the switch
    statement into the style where the case statements are at the same
    level.  When comparing with older source, one may use the -b argument
    of the diff command to ignore changes in amount of white space:

    	diff -b -r -u calc-2.11.11 calc-2.12.0

    The read, write, and help commands use the value of global string
    variable if the symbol name starts with a $.  For example:

    	global x = "lucas.cal";
	read $x;	/* same as read lucas.cal or read "lucas.cal" */

    Added dotest.cal resource.  Based on a design by Ernest Bowen
    <ebowen at une dot edu dot au>, the dotest evaluates individual
    lines from a file.  The dotest() function takes 1 to 3 arguments:

	dotest(dotest_file [,dotest_code [,dotest_maxcond]])

	dotest_file

	    Search along CALCPATH for dotest_file, which contains lines that
	    should evaluate to 1.  Comment lines and empty lines are ignored.
	    Comment lines should use ## instead of the multi like /* ... */
	    because lines are evaluated one line at a time.

	dotest_code

	    Assign the code number that is to be printed at the start of
	    each non-error line and after **** in each error line.
	    The default code number is 999.

	dotest_maxcond

	    The maximum number of error conditions that may be detected.
	    An error condition is not a sign of a problem, in some cases
	    a line deliberately forces an error condition.  A value of -1,
	    the default, implies a maximum of 2147483647.

	Global variables and functions must be declared ahead of time because
	the dotest scope of evaluation is a line at a time.  For example:

	    ; read dotest.cal
	    ; read set8700.cal
	    ; dotest("set8700.line");

    Updated the todo / wish list items.  The top priority now is to
    convert calc to GNU autoconf / configure to build the calc.

	; help todo

    Added missing help file for the stoponerror() builtin.

    Corrected and improved the help documentation for factor and lfactor.

    Fixed a problem where some error messages that should have been
    written to a file or string, went to stderr instead.  This bug was
    fixed by Ernest Bowen <ebowen at une dot edu dot au>.

    Corrected the documentation relating to the calc -c command line option.
    The -c relates to scan/parse errors only, not execution errors.

    Corrected a stack overflow problem where the math_fmt() in zio.c
    could be induced to overflow the stack.  This problem was independently
    reported by Chew Keong Tan of Secunia Research <vuln at secunia dot com>.

    Corrected a stack overflow problem where the scanerror() in token.c
    could be induced to overflow the stack by a malformed token.

    Made math_error() in math_error.c more robust against a error
    message constant that is too long.

    Made read_bindings() in hist.c more robust against very long bindings
    config lines.

    Made listsort() in listfunc.c and matsort() matfunc.c more robust
    against sorting of impossibly huge lists and matrices.

    Warnings about an undefining a builtin or undefined function, a
    constant before the comma operator, and an unterminated comment is
    now processed by scanerrors (not simply written directly to stderr).
    These warnings file and line number in which the "error" occurred
    as well as a more precise message than before.  If using -c on the
    calc command line or if stoponerror(-1), then assuming there are
    no other compile errors, only the unterminated comment will stop
    completion of the function being defined.

    The cal/regress.cal now reads most of the calc resource files.

    The issq() test had a slight performance boost.  A minor note
    was added to the help/issq file.

    Improved the documentation of the mul2, sq2, pow2, and redc2 config
    parameters in help/config.

    Added config("baseb"), a read-only configuration value to return
    the number of bits in the fundamental base in which calculations
    are performed.  This is a read-only configuration value.

    Calc now will allow syntax such as ++*p-- and  ++*----*++p----
    where p is an lvalue; successful evaluation of course require the
    successive operations to be performed to have operands of appropriate
    types; e.g. in *A, A is usually an lvalue whose current value is a
    pointer. ++ and -- act on lvalues. In the above examples there are
    implied parentheses from the beginning to immediately after p. If
    there are no pre ++ or -- operations, as in **p++.  The implied
    parentheses are from immediately before p to the end.

    Improved the error message when && is used as a prefix operator.

    Changed the help/config file to read like a builtin function help file.

    One can no longer set to 1, or to a value < 0, the config()
    parameters: "mul2", "sq2", "pow2", and "redc2".  These values
    in the past would result in improper configuration of internal
    calc algorithms.  Changed cal/test4100.cal to use the minimal
    value of 2 for "pow2", and "redc2".

    Changed the default values for the following config() parameters:

    	config("mul2") == 1780
	config("sq2") == 3388
	config("pow2") == 176

	These values were determined established on a 1.8GHz AMD 32-bit
	CPU of ~3406 BogoMIPS by the new resource file:

	    cal/alg_config.cal

   Regarding the alg_config.cal resource file:

	The best_mul2() function returns the optimal value of config("mul2").
	The best_sq2() function returns the optimal value of config("sq2").
	The best_pow2() function returns the optimal value of config("pow2").
	The other functions are just support functions.

	By design, best_mul2(), best_sq2(), and best_pow2() take a few
	minutes to run.  These functions increase the number of times a
	given computational loop is executed until a minimum amount of CPU
	time is consumed.  To watch these functions progress, one can set
	the config("user_debug") value.

	Here is a suggested way to use the alg_config.cal resource file:

	    ; read alg_config
	    ; config("user_debug",2),;
	    ; best_mul2(); best_sq2(); best_pow2();
	    ; best_mul2(); best_sq2(); best_pow2();
	    ; best_mul2(); best_sq2(); best_pow2();

	NOTE: It is perfectly normal for the optimal value returned
	to differ slightly from run to run.  Slight variations due to
	inaccuracy in CPU timings will cause the best value returned to
	differ slightly from run to run.

	See "help resource" for more information on alg_config.cal.

    Updated the "help variable" text to reflect the current calc
    use of ` (backquote), * (star), and & (ampersand).

    Removal of some restrictions on the use of the same identifier
    for more than one of parameter, local, static or global variable.

	For example, at command level, one could use:

	    for (local x = 0; x < 10; x++) print sqrt(x);

	At the beginning of a statement, "(global A)" is a way of
	indicating a reference to the variable A, whereas "global A"
	would be taken as a declaration. Parentheses are not required in
	"++global A" or "global A++" when "global" is used in this way.

	The patch extends this "specifier" (or "qualifier") feature
	to static variables, but such that "static A" refers only
	to a static variable at the current file and function scope
	levels. (If there is already a static variable A at the current
	file and function levels, a declaration statement "static A"
	would end the scope of that variable and define a new static
	variable with identifier A. A "global A" declaration is more
	drastic in that it ends the scope of any static variable A at
	the same or higher scope levels.)

	Unlike a static declaration in which an "initialization" occurs at
	most once, in the specifier case, "static A = expr" is simply an
	assignment which may be repeated any number of times.  An example
	of its use is:

	    define np() = static a = nextprime(a);

	For n not too large, the n-th call to this function will
	return the n-th prime. The variable a here will be private to
	the function.

	Because one can use "global", "local" or "static" to specify a
	type of variable, there seems little point in restricting the
	ways identifiers that can be used in more than one of these
	or as parameters. Obviously, introducing A as a local variable
	when it is being used as a parameter can lead to confusion and a
	warning is appropriate, but if it is to be used only occasionally,
	it might be convenient to be able to refer to it as "local A"
	rather than introducing another identifier. While it may be
	silly to use the same identifier for both a parameter and local
	variable, it should not be illegal.

    Added warnings for possibly questionable code in function definitions.

    Added config("redecl_warn", boolean) to control if calc issues
    warnings about variables being declared.  The config("redecl_warn")
    value is TRUE by default.

    Added config("dupvar_warn", boolean) to control if calc issues
    warnings about when variable names collide.  The config("dupvar_warn")
    value is TRUE by default.  Examples of variable name collisions
    include when:

    	* both local and static variables have the same name
    	* both local and global variables have the same name
    	* both function parameter and local variables have the same name
    	* both function parameter and global variables have the same name

    Fix of a bug which causes some static variables not to be correctly
    unscoped when their identifiers are used in a global declaration.

    Change of "undefine" from a command-level keyword to statement level and
    introduction of an "undefine static A" statement to end the scope of a
    static variable A at the current file/function levels.

    Change/restored the syntax rules for "for" and "while" loops to
    recognize an unescaped newline in top-level command-level statements.

    Updated help/avg, help/define, help/fprintf, help/gcd, help/hash,
    help/hmean, help/lcm, help/max, help/min, help/null, help/poly,
    help/printf, help/ssq, help/strcat, help/strprintf, help/sum,
    help/xor.

    Changed the definition of the function ssq() to enable list arguments
    to be processed in the same way as in sum().  For example:

    	ssq(1,2, list(3,4,list(5,6)), list(), 7, 8)

    returns the value of 1^2 + 2^2 + ... + 8^2 == 204.

    Added the calc resource sumtimes.cal, to give the runtimes for
    various ways of evaluating sums, sums of squares, etc, for large
    lists and matrices.  For example:

    	read sumtimes
	doalltimes(1e6)

    Calc now ignores carriage returns (\r), vertical tabs (\v), and
    form feeds (\f) when token parsing.  Thus users on Windoz systems
    can write files using their \r\n format and users on non-Windoz
    systems can read them without errors.

    The quomod() builtin function now takes an optional 5th argument
    which controls the rounding mode like config("quomod") does, but
    only for that call.  Now quomod() is in line with quo() and mod()
    in that the final argument is an optional rounding mode.

    Added a "make uninstall" rule which will attempt to remove everything
    that was installed by a "make install".

    Changed the "Copyright" line in the rpm spec file to a "License" line
    as per new rpm v4.4 syntax.

    The quomod() builtin function does not allow constants for its 3rd
    and 4th arguments.  Updated the "help quomod" file and added more
    quomod regression tests.

    Added patch from Ernest Bowen <ebowen at une dot edu dot au> to
    add the builtin: estr().  The estr(x) will return a representation
    of a null, string, real number, complex number, list, matrix,
    object. block, named block, error as a string.

    Added patch from Ernest Bowen <ebowen at une dot edu dot au> to
    add the builtin: fgetfile().  The fgetfile(x) will return the rest
    of an open file as a string.

    Improved help files for fgetfield, fputs, name, or quomod.


The following are the changes from calc version 2.11.10.1 to 2.11.11:

    Fixed a bug reported by the sourceforge user: cedars where:

    	ln(exp(6)) == 3		/* WRONG!!! */

    incorrectly returned 1.  This bug was fixed by Ernest Bowen
    <ebowen at une dot edu dot au>.  The regression test
    was expanded to cover this issue.

    Added minor improvements to hash regression testing of pi().

    Fixed "help script" and the calc man page regarding the requirement
    of -f to be the last -flag in shell script mode.  Further clarified
    the meaning and placement of the -f flag.

    Moved issues with chi.cal intfile.cal into a "mis-features" section
    of the BUGS file.  See "help bugs" or the BUGS source file for details.

    Added the bug about:

	calc 'read ellip; efactor(13*17*19)'

    to the BUGS file.  See "help bugs" or the BUGS source file for details.
    Anyone want to track down and fix this bug?

    Fixed typo in the "help mat" example and improved the mat_print example.

    Renamed most COMPLEX C function names to start with c_ to avoid
    conflicts with new C standard functions.  Note that the calc
    builtin function names remain the same.   The C function names
    inside the C source that calc is written in changed.  This means
    that code that linked to libcalc.a will need to change in order
    to call calc's functions instead of the C standard functions.
    See cmath.h, comfunc.c, and commath.c for details.  See also
    http://www.opengroup.org/onlinepubs/009695399/basedefs/complex.h.html
    for names of the new C standard functions.

    Changed the calc man page to note that using -- in the command will
    separate calc options from arguments as in:

	calc -p -- -1 - -7

    Noted how Apple OS X can make use of readline in the Makefile.
    In particular:

	# For Apple OS X: install fink from http://fink.sourceforge.net
	#                 and then do a 'fink install readline' and then use:
	#
	READLINE_LIB= -L/sw/lib -lreadline -lhistory -lncurses

    Added linear.cal as a calc standard resource file.


The following are the changes from calc version 2.11.10 to 2.11.10:

    The cygwin config value is correctly tested while doing comparisons
    between config states.

    Added config("compile_custom") to determine if calc was compiled
    with -DCUSTOM.  By default, the Makefile uses ALLOW_CUSTOM= -DCUSTOM
    so by default, config("compile_custom") is TRUE.  If, however,
    calc is compiled without -DCUSTOM, then config("compile_custom")
    will be FALSE.  NOTE: The config("compile_custom") value is only
    affected by compile flags.  The calc -D runtime command line option
    does not change the config("compile_custom") value.  This is a
    read-only configuration value.

    Added config("allow_custom") to determine if the use of custom
    functions are allowed.  To allow the use of custom functions, calc
    must be compiled with -DCUSTOM (which it is by default) AND calc run
    be run with the -D runtime command line option (which it is not by
    default).  If config("allow_custom") is TRUE, then custom functions
    are allowed.  If config("allow_custom") is FALSE, then custom
    functions are not allowed.  This is a read-only configuration value.

    Correctly hash config state for windows and cygwin values.  The value
    of config("compile_custom") and config("allow_custom") also affect
    the hash of the config state.

    Fixed the custom/argv.cal test code to avoid use of a reserved
    builtin function name.

    Fixed custom/*.cal scripts to conform better with the cal/*.cal
    resource files.

    Removed the Makefile variables ${LONGLONG_BITS}, ${HAVE_LONGLONG},
    and ${L64_FORMAT}.  Removed longlong.c and longlong.h.  The use
    of HAVE_LONGLONG=0 was problematic.  The lack of complaints about
    the HAVE_LONGLONG=0 shows that the 'long long' type is wide spread
    enough warrent not trying to support compilers without 'long long'.

    Removed the SVAL and SHVAL macros from zrand.c, zrand.h, and zmath.h
    as they were causing too many broken C pre-processors and C checkers
    to become confused.

    Added a 'make splint' rule to use the splint statically checking
    tool on the calc source.

    Removed support of the BSDI platform.  The BSDI platform is no longer
    directly supported and we lost our last BSDI machine on which we
    could test calc.  Best wishes to the former BSDI folk and thanks
    for breaking important ground in the Open Source Movement!

    Fixed several typos found in the documentation and buildin
    function output by C Smith <smichr at hotmail dot com>.

    Fixed -d so that:

    	calc -d 2/3

    will print 0.66666666666666666667 without the leading tilde as
    advertised in the man page.

    Added a missing help file for the display builtin function as
    requested by Igor Furlan <primorec at sbcglobal dot net>.

    Changed the "help environment" file to reflect modern default
    values of CALCPATH and CALCRC.

    Added missing variables for printing by the "make env" rule.

    Added EXT Makefile variable so that Cygwin can install calc as
    calc.exe.  By default, EXT is empty so that calc is calc on most
    modern operating systems.  Thanks goes to Ullal Devappa Kini <wmbfqj
    at vsnl dot net> for helping identify this problem and testing our fix.

    Added custom function:

    	custom("pmodm127", q)

    to compute 2^(2^127-1) mod q.  While currently slower than just
    doing pmod(2,2^127-1,q), it is added to give an example of a
    more complex custom function.  Call calc with the -C flag to
    use custom functions.

    Made slight changes to the custom/HOW_TO_ADD documentation.

    Fixed some \ formatting man page problems as reported by Keh-Cheng
    Chu <kehcheng at quake dot Stanford dot edu>.

    Fixed some comparison between signed and unsigned in md5.c
    that was reported for the PowerMac G5 2GHz MacOS 10.3 by
    Guillaume VERGNAUD <vergnaud at via dot ecp dot fr>.

    Fixed a number of pending issues with help files filling in
    missing  LIMITS, LINK LIBRARY, and SEE ALSO information,


The following are the changes from calc version 2.11.9 to 2.11.9.3:

    Fixed calc man page examples to move -f to the end of the line.
    Thanks goes to Michael Somos <somos at grail dot cba dot csuohio
    dot edu> for pointing this out.

    Linux and gcc now compiled with -Wall -W -Wno-comment.

    Fixed a post increment that was reported by R. Trinler <trinler at
    web dot de> and fixed by Ernest Bowen <ernie at turing dot une dot
    edu dot au>.

    Fixed pi.cal to not depend on the buggy pre-2.11.9 post increment
    behavior.

    Added config("cygwin") to determine if calc was compiled under Cygwin.
    The config("cygwin") is a read-only configuration value that is 1
    when calc was compiled under Cygwin and 0 otherwise.  Regression
    tests 949 and 950 are skipped when config("cygwin") is true.

    The Makefile variable HAVE_NO_IMPLICIT is empty by default so that
    the Makefile will test if the compiler has a -Wno-implicit flag.

    Added HAVE_UNUSED Makefile variable.  If HAVE_UNUSED is empty,
    then the Makefile will run the have_unused program to determine
    if the unused attribute is supported.  If HAVE_UNUSED is set to
    -DHAVE_NO_UNUSED, then the unused attribute will not be used.

    The Makefile builds have_unused.h which defines, if the unused
    attribute is supported:

	#define HAVE_UNUSED /* yes */
	#define UNUSED __attribute__((unused)) /* yes */

    or defines, if the unused is not supported (or if the Makefile
    variable is HAVE_UNUSED= -DHAVE_NO_UNUSED):

	#undef HAVE_UNUSED /* no */
	#define UNUSED /* no */

    Fixed numerous warnings about comparison between signed and unsigned
    value warnings and unused parameter warnings in version.c, zrand.c,
    string.c, shs1.c, shs.c, qtrans.c, qmath.c, qfunc.c, md5.c, matfunc.c,
    hist.c, file.c, const.c, blkcpy.c, seed.c, opcodes.c, func.c, qio.c,
    zrandom.c, custom/c_argv.c, custom/c_devnull.c, custom/c_help.c,
    custom/c_sysinfo.c, addop.c and calc.c.

    Fixed some typos in this file.

    By default, compile with -O3 -g3.  The Makefile comments on how some
    distributions might need to use -O2 -g or -O -g.


The following are the changes from calc version 2.11.8.0 to 2.11.8.1:

    Updated HOWTO.INSTALL to reflect the new RPM files.

    Clarify that the internal hash as well as the hash builtin
    function used by calc, while based on the Fowler/Noll/Vo
    hash is NOT an FNV hash.

    Made slight performance improvements to calc by an optimization of how
    calc's internal hash is computed.  The "make chk" regression test
    runs about 1.5% faster (when compiled with -O3 on an AMD Athlon)
    NO_HASH_CPU_OPTIMIZATION is not defined.  Calc's internal hash values
    have not changed.  By default, NO_HASH_CPU_OPTIMIZATION is NOT defined
    and the slightly faster expression is used.

    A slight modification of what was known as the "calc new standard"
    configuration (calc -n or config("all", "newstd")) is now the default
    calc configuration.  The flag:

    	calc -O

    was added to get the old classic calc configuration.  The flag command
    line flag, -n, now does nothing.  Use of -n is deprecated and may go
    away / be used for something else in the future.

    The following table gives the summary of these changes:

	     pre v2.11.8		     v2.11.8
	     default         pre v2.11.8     -O & oldstd      v2.11.8
	     and oldstd	     -n & newstd     classic cfg      default
	     --------------------------------------------------------
    epsilon	1e-20		1e-10		1e-20		1e-20
    quo	    	2		2		2		2
    outround	2		24		2		24
    leadzero	0		1		0		1
    fullzero	0		1		0		0
    prompt	>		;		>		;
    more	>>		;;		>>		;;

    With the exception of epsilon being 1e-20, and fullzero being unset,
    the new default calc config is like it was (pre-2.11.8) with calc -n /
    config("all", "newstd").

    The new default config is the old classic config with outround being
    24, leadzero being set, and the prompts being ;'s.

    Fixed a bug in the evaluation of tanh(1e-23) with an epsilon(1e-100).
    Thanks goes to Dmitry G. Baksheyev <bd at nbsp dot nsk dot su>
    for reporting the problem, and thanks goes to Ernest Bowen
    <ernie at turing dot une dot edu dot au> for the fix.


The following are the changes from calc version 2.11.7.0 to 2.11.7.1:

    Added support to build calc RPMs thanks to Petteri Kettunen
    <petterik at users dot sourceforge dot net>.

    Added rpm rule to Makefile to build rpm set.  The rpm rule
    uses the rpm.mk Makefile and the calc.spec.in spec template.

    The default Makefile is now the Makefile used during rpm
    creation.  This Makefile assumes that system has readline,
    ncurses (-lreadline -lhistory -lncurses), and less.
    It compiled with a high gcc optimization level (-O3 -g3).
    The Makefile used during rpm creation is the Makefile
    that appears in the calc-src rpm as well.

    The Makefile shipped with the old style gziped tarball
    is still the same generic Makefile.

    The Makefile now uses ${MKDIR} ${MKDIR_ARG} when creating
    directories during installation.  By default, it does
    a mkdir -p when forming directories.

    Fixed attributes on include and lib calc-devel files.

    Adjusted the interaction between rpm.mk, and the calc.spec.in.
    Release number now comes from calc.spec.in only.

    Renamed calc and calc-devel RPMs to use .i686 instead of .i386.


The following are the changes from calc version 2.11.6.3 to date:

    Fixed a bug in deg.cal where fixdms() was being called with
    the wrong type of argument.

    Changed the value of digits(1) and digits(0) to be 1.  Now digits()
    returns number of digits in the standard base-b representation
    when x is truncated to an integer and the sign is ignored.
    To be more precise: when abs(int(x)) > 0, this function returns
    the value 1 + ilog(x, b).  When abs(int(x)) == 0, then this
    function returns the value 1.

    As the result of the above digits() change, the repeat.cal
    resource file script was modified to remove the special
    case for repeating a value of 1.  Also the regress tests
    #715, #977 and #978 were changed.

    Made a minor improvement to the "help places" documentation.

    Fixed dms_neg(a) in deg.cal thanks to a bug report by kaa
    <kaa76 at pochtamt dot ru>.


The following are the changes from calc version 2.11.6.0 to 2.11.6.2:

    Clarified remark in lucas.cal about use of n mod 2^n == 0.

    Fixed help typos reported by Marc Mezzarobba <mm at mm dot ovh dot org>.

    Forced system("") to return 0 under windoz.

    The direct.h include file is not used when compiling under Cygwin.

    Fixed bug where random(10,11) caused calc to dump core when issued
    the 2nd time.

    Moved the setting of the Makefile variable ${CALC_INCDIR} to
    the section where things like ${BINDIR} and ${LIBDIR} are set.
    Idea from Clifford Kite <kite_public1 at ev1 dot net>.

    The Makefile is shipped mode 0644 since a number of folks
    edit it (to build and check calc) as a non-root user and later
    on su to root to install.  Idea from Clifford Kite <kite_public1
    at ev1 dot net>.

    Added base2() builtin function to calc.  Normally calc prints
    values according to base().  Frequently some users want to see
    a value in two bases.  Flipping back and forth between to bases
    is a bit of a pain.  With base2(), calc will output a value twice:

	; 234567
		234567
	; base2(16),
	; 234567
		234567 /* 0x39447 */
	; 131072
		131072 /* 0x20000 */
	; base2(0),
	; 131072
        131072

    By default, base2() is disabled.  Calling base2(0) will also turn
    off the double base mode.  Thanks goes to Erik Anggard
    <erik dot anggard at packetfront dot com> for his idea and
    his initial patch.

    Added repeat.cal as a calc resource file script:

	repeat(digit_set, repeat_count)

	Return the value of the digit_set repeated repeat_count times.
	Both digit_set and repeat_count must be integers > 0.

	For example repeat(423,5) returns the value 423423423423423,
	which is the digit_set 423 repeated 5 times.

    Makefile no longer makes a direct reference to Red Hat 6.0.

    Added missing math_setmode2() prototype to zmath.h.

    Fixed some implicit declarations of functions by either making
    them explicit or by including the proper system .h files.

    Makefile no longer uses -Wno-implicit flag, by default, for
    gcc based compiles on calc source.  Makefile now attempts to
    compile no_implicit.c with an explicit -Wno-implicit arg in an
    effort to determine of -Wno-implicit is a valid compiler flag.
    If no_implicit.c is compiled with -Wno-implicit, then
    the file no_implicit.arg is created with the contents
    of the -Wno-implicit flag.  Otherwise no_implicit.arg
    is created as an empty file.

    Added the Makefile variable ${HAVE_NO_IMPLICIT}, which if
    not set to YES will prevent no_implicit.c from being
    compiled and prevent the -Wno-implicit flag from being used.
    If ${HAVE_NO_IMPLICIT} is not YES, then an empty no_implicit.arg
    file is created and no_implicit.c is not compiled.

    The seed.c file, because the pseudo_seed() function contains
    calls to a number of various system functions, attempts to
    compile with the -Wno-implicit flag (if allowed by the
    formation of the no_implicit.arg file).

    Misc make depend fixes and cleanup.

    Fixed formation of the custom/.all file.

    Fixed repeat(1, repeat_count) bug.


The following are the changes from calc version 2.11.5.5 to 2.11.5.9:

    Now using version numbers of one of these forms:

    	x.y.z.w
    	x.y.z
    	x.y

    Changed the READLINE_LIB Makefile variable to not link with -lreadline
    by default.  If you do have readline, we recommend that you use it.
    If you can install the GNU readline:

	http://freshmeat.net/projects/gnureadline/
    	http://cnswww.cns.cwru.edu/php/chet/readline/rltop.html

    we recommend it.  But if not, you should set the USE_READLINE,
    READLINE_LIB, and READLINE_INCLUDE Makefile variables to empty.
    NOTE: See the BUGS file for a Linux issue when compiling calc
    with -O (or -O2 or -O3) AND with -g (or -g3) AND with readline.

    Removed an obsolete reference to TOPDIR.  This was fixed thanks to
    a bug report by Clifford Kite <kite_public1 at ev1 dot net>.
    Fixed other inconsistencies related to things like BINDIR.

    Fixed calc man page so that is refers to -f instead of the old -S flag.
    Fixed thanks to Clifford Kite <kite_public1 at ev1 dot net> for
    point this out.

    All for loops end with /dev/null to avoid any problems related
    to systems that cannot grok empty for loops.

    Changed the libcalc functions creal and cimag to c_real and c_imag
    to about conflicts with new libc such as those used by gcc v3.
    Thanks Eli Zaretskii <eliz at is dot elta dot co dot il> and
    Martin Buck <m at rtin-buck dot de> for alerting us to this conflict.

    The Makefile no longer hard code's /usr/include.  Instead it uses
    the ${INCDIR} Makefile variable.  Thanks goes to Eli Zaretskii
    <eliz at is dot elta dot co dot il> for pointing out this inconsistency.

    Added mods to support compilation under DJGPP.  DJGPP runs on 386
    and newer PCs running DOS or dos-compatible operating systems.
    See http://www.delorie.com/djgpp/.  Thanks goes to Eli Zaretskii
    <eliz at is dot elta dot co dot il> for sending in these mods.

    Updated README.WINDOWS to include information on building with DJGPP.

    The pld folks are building RPMs based on our calc distributions.
    See:  ftp://ftp.pld.org.pl/dists/ra/PLD/i686/PLD/RPMS or
    http://ftp.pld.org.pl/dists/ra/PLD/i686/PLD/RPMS more information.
    We appreciate their work in this regard.  In the next release, we
    plan to also build and release our own RPMs based on their efforts.

    Changed the Makefile variable CUSTOMLIBDIR to CUSTOMCALDIR.
    Changed the Makefile variable CSHAREDIR to CALC_SHAREDIR.
    Changed the Makefile variable INCDIRCALC to CALC_INCDIR.
    Removed the Makefile variable SHAREDIR.

    Updated the HOWTO.INSTALL and README.WINDOWS files.

    Fixed definition of MAXUFULL.  Thanks to a bus report from
    Jill Poland <jpoland at cadence dot com>.

The following are the changes from calc version 2.11.5t4.1 to 2.11.5t4.4:

    Updated dependency rules in Makefiles.

    NOTE: -DSRC, as used in 2.11.5t4.1 was renamed -DCALC_SRC
    in a later version.

    Calc include files use #include "foo.h" to include other calc
    header files if -DCALC_SRC.  Otherwise they use <calc/foo.h>.
    The -DCALC_SRC symbol is defined by default in calc's Makefile
    and so it uses the header files from within the calc src tree.
    If an external non-calc program includes an installed calc
    header file (from under /usr/include), and it does NOT define
    CALC_SRC, then it will obtain the calc header files from the
    correct system location (such as /usr/include/calc/foo.h).

    Added calc builtin function: version() which returns the calc
    version string.

    Added subject requirements for the calc-tester-request and
    calc-bugs-mail EMail aliases.  See:

    	http://www.isthe.com/chongo/tech/comp/calc/email.html

    for details.

    Corrected a bug that incorrectly set the default calc path
    back in version 2.11.5t4.  The default CALCPATH is now:

	.:./cal:~/.cal:/usr/share/calc:/usr/share/calc/custom

    and the default CALCRC is now:

	/usr/share/calc/startup:~/.calcrc:./.calcinit

    This fixes the missing bindings error and it places the calc
    resource files into the default path.

    If you are using the GNU readline then the Makefile recommends that
    you link with the ncurses library.

    Applied Makefile, cscript/Makefile and custom/Makefile patches to
    fix install mode problems, to deal with sorting and dates in I18n
    environments (such as Japanese), to fix some problems with calc.spec
    and to fix the cscript #! header lines.  Thanks goes to KAWAMURA Masao
    (kawamura at mlb.co.jp) for the bug report and patch!

    Fixed headers on fproduct.calc powerterm.calc 4dsphere.calc so
    that they are correcly changed on installation.

    Added ${GREP} Makefile variable.

    The top level Makefile now sets LANG=C and passes it down to
    lower level Makefiles.

    Updated URLs in cal/lucas.cal comments.

    Now shipping calc.spec, inst_files, spec-template and Makefile.linux
    with the standard calc source distribution.  Note that the standard
    Makefile has not changed.  The Makefile.linux only in minor ways
    needed to build calc rpms.

    Added $T Makefile variable.  $T is the top level directory under
    which calc will be installed.  The calc install is performed under $T,
    the calc build is performed under /.  The purpose for $T is to allow
    someone to install calc somewhere other than into the system area.
    For example when forming the calc rpm, the Makefile is called with
    T=$RPM_BUILD_ROOT.  If $T is empty, calc is installed under /.

    Removed all echo_XYZ rules except for echo_inst_files from lower
    level makefile.  The calc.spec will use a make install rule
    with T=$RPM_BUILD_ROOT.

    Updated LIBRARY file with instructions related to -DCALC_SRC,
    the new default include file location and -lcustcalc.


The following are the changes from calc version 2.11.5t3 to 2.11.5t4:

    The Makefile will now send both stdout and stderr to /dev/null
    when compiling hsrc intermediates.

    The config("verbose_quit") value was restored to a default
    value of FALSE.

    Added the cscript:

	powerterm [base_limit] value

    to write the value as the sum (or difference) of powers <= base_limit
    where base_limit by default is 10000.

    Applied a bug fix by Dr.D.J.Picton <dave at aps5.ph.bham.ac.uk>
    to have help with no args print the default help file.

    Renamed lavarand to LavaRnd.

    Added rules to build a calc rpm.

    All installed files are first formed as foo.new, and then moved
    into place as foo via a atomic rename.

    During installation, only files that are different are installed.
    If the built file and the installed file are the same, no
    installation is performed.

    Calc has new default installation locations:

    Makefile var   old location			      new location
    ------------   ------------			      ------------
    TOPDIR	   /usr/local/lib		        <<no longer used>>
    BINDIR	   /usr/local/bin		      /usr/bin
    SHAREDIR	      <<not set>>		      /usr/share
    INCDIR	   /usr/local/include		      /usr/include
    LIBDIR	   /usr/local/lib/calc		      /usr/lib
    CSHAREDIR	      <<not set>>		      /usr/share/calc
    HELPDIR	   /usr/local/lib/calc/help           /usr/share/calc/help
    INCDIRCALC	   /usr/local/include/calc	      /usr/include/calc
    CUSTOMLIBDIR   /usr/local/lib/calc/custom	      /usr/share/calc/custom
    CUSTOMHELPDIR  /usr/local/lib/calc/help/custhelp  /usr/share/calc/custhelp
    CUSTOMINCDIR     <<not set>>		      /usr/include/calc/custom
    SCRIPTDIR	   /usr/local/bin/cscript	      /usr/bin/cscript
    MANDIR	     <<not set>>		      /usr/share/man/man1
    CATDIR	     <<not set>>		        <<not set>>

    The Makefile variable ${TOPDIR} is no longer used.  In some places
    it has been replaced by a new Makefile variable ${SHAREDIR}.  Some
    of the old TOPDIR functionality has been replaced by ${CSHAREDIR}.

    The install rules no longer remove old obsolete files.  We assume
    that these old files have long since vanished!  :-)

    Reduced the amount of output when doing a make all where nothing
    needs to be made.

    Reduced the amount of output when doing a make install where nothing
    needs to be installed.

    If you install using the new default locations, you can remove
    old calc files installed in the old default location by doing:

	make olduninstall


The following are the changes from calc version 2.11.5t2 to 2.11.5t2.1:

    Fixed a bug, reported by Ernest Bowen <ernie at turing dot
    une dot edu dot au> that caused command lines to be echoed in
    interactive mode.  Fixed a bug that sometimes left the terminal
    in a non-echoing state when calc exited.

    Renamed error codes E_FGETWORD1 and E_FGETWORD2 symbols to
    E_FGETFIELD1 and E_FGETFIELD2.

    Made a minor format change to the top of the calc man page.

    The findid() function in file.c 2nd argument changed.  The argument
    is now mostly a writable flag.  This function now finds the file
    I/O structure for the specified file id, and verifies that
    it is opened in the required manner (0 for reading or 1 for writing).
    If the 2nd argument is -1, then no open checks are made at all and
    NULL is then returned if the id represents a closed file.

    The calc builtin function, fopen(), now allows one to specify
    opening files in binary modes.  On POSIX / Linux / Un*x-like systems,
    text file is the same as a binary file and so 'b' to an fopen has
    no effect and is ignored.  However on systems such as MS Windoz
    the 'b' / binary mode has meaning.  See 'help fopen' for details.

    On systems (such as MS Windoz), calc will produce a different error
    message when it attempts to open /dev/tty.  This will condition
    will occur in things like calc scripts when they switch from ``batch
    processing'' commands from and want to start interactive mode.

    Regression tests fopen in binary mode in a few places where a
    difference between text and binary string lengths matter.
    The intfile calc resource file also uses binary mode.

    Changed the rand() builtin and its related functions srand() and
    randbit() to use the Subtractive 100 generator instead of the
    additive 55 generator.  This generator as improved random properties.
    As a result, of this change, the values produced by rand(),
    rand() and randbit() are now different.

    Updated regression tests for new rand() and randbit() output.

    Applied a bug fix from Ernest Bowen <ernie at turing dot une dot
    edu dot au> dealing with one-line "static" declaration like:

    	static a = 1, b;

    Added regression test 8310 to test for the static bug fix.


The following are the changes from calc version 2.11.5t0 to 2.11.5t1.1:

    Fixed a compile problem with Linux 2.4 / Debian.  Thanks goes
    to Martin Buck <m at rtin-buck dot de> for help with this issue.

    Fixed a bug in how L64_FORMAT (it determined if "%ld" or "%lld"
    is appropriate for printing of 64 bit long longs) was determined.
    Thanks goes to Martin Buck <m at rtin-buck dot de> for reporting
    this bug and testing the fix.

    An effort was made to make calc easier to build under Windoz
    using the Cygwin project (http://sources.redhat.com/cygwin/).
    Thanks to the work of Thomas Jones-Low (tjoneslo at softstart
    dot com), a number of #if defined(_WIN32)'s have been added
    to calc source.  These changes should not effect Windoz
    free system such as GNU/Linux, Solaris, POSIX-like, etc ...

    Added windll.h to deal with Windoz related DLL issues.
    Using the convention of 'extern DLL' instead of 'DLL extern'
    to deal with symbols that export to or import from a DLL.

    Added HAVE_MALLOC_H, HAVE_STDLIB_H, HAVE_STRING_H, HAVE_TIMES_H,
    HAVE_SYS_TIMES_H, HAVE_TIME_H, HAVE_SYS_TIME_H, HAVE_UNISTD_H
    and HAVE_URANDOM to the Makefile.  If these symbols are empty,
    then the Makefile looks for the appropriate system include file.
    If they are YES, then the Makefile will assume they exist.
    If they are NO, then the Makefile will assume they do not exist.

    Changed HAVE_URANDOM to match the empty, YES, NO values.
    If HAVE_URANDOM is empty, then the Makefile will look for /dev/urandom.
    If HAVE_URANDOM is YES, then the Makefile will assume /dev/urandom exists.
    If HAVE_URANDOM is NO, then the Makefile will assume /dev/urandom does
    not exist.

    If TERMCONTROL is -DUSE_WIN32, then the Windoz terminal control
    (no TERMIOS, no TERMIO, no SGTTY) will be assumed.

    Added a win32_hsrc Makefile rule to create hsrc files appropriate
    for a Windoz system using Cygwin gcc environment.  Added win32.mkdef
    which is used by the win32_hsrc rule to set the Windoz specific
    Makefile values to build hsrc files.  The hsrc files are built
    under the win32 directory.

    Added FPOS_POS_BITS, OFF_T_BITS, DEV_BITS and INODE_BITS Makefile
    symbols to allow one to force the size of a file position, file
    offset, dev and inode value.  Leaving these values blank will
    Makefile to determine their size.

    Fixed a bug in the way file offsets, device and inode values are copied.

    Added chi.cal for a initial stab as a Chi^2 function.  The chi_prob()
    function does not work well with odd degrees of freedom, however.

    Added big 3 to config("resource_debug").  Calc resource file scripts
    check for config("resource_debug") & 8 prior to printing internal debug
    statements.  Thus by default they do not print them.

    Added intfile.cal as a calc resource file script:

	file2be(filename)

	    Read filename and return an integer that is built from the
	    octets in that file in Big Endian order.  The first octets
	    of the file become the most significant bits of the integer.

	file2le(filename)

	    Read filename and return an integer that is built from the
	    octets in that file in Little Endian order.  The first octets
	    of the file become the most significant bits of the integer.

	be2file(v, filename)

	    Write the absolute value of v into filename in Big Endian order.
	    The v argument must be on integer.  The most significant bits
	    of the integer become the first octets of the file.

	le2file(v, filename)

	    Write the absolute value of v into filename in Little Endian order.
	    The v argument must be on integer.  The least significant bits
	    of the integer become the last octets of the file.

    Added the following help aliases:

	copy	blkcpy
	read	command
	write	command
	quit	command
	exit	command
	abort	command
	cd	command
	show	command

    Added the cscript:

	fproduct filename term ...

    to write the big Endian product of terms to a filename.  Use - for stdout.

    Fixed calc path in help/script.

    Added read-only parameter, config("windows") to indicate if the system
    is MS windowz WIN32 like system.

    Configuration values that used to return "true" or "false" now return
    1 (a true value) or 0 (a false value).  Thus one can do:

    	if (config("tab")) { ... } else { ... }

    The configuration values that now return 1 or 0 are:

	config("tilde")
	config("tab")
	config("leadzero")
	config("blkverbose")
	config("verbose_quit")
	config("windows")

    Now shipping a win32 sub-directory that contains hsrc .h files
    that have been attempted to be built for windoz.


The following are the changes from calc version 2.11.4t1 to 2.11.4t2:

    Added missing test8600.cal test file.

    Fixes cscript files to deal with the -S flag being replaced by
    -f and possibly other flags.

    Added regression tests for builtin functions bernoulli, catalan,
    euler, freeeuler, and sleep.  Added non-base 10 regression tests
    for digit, digits and places.

    The bernoulli.cal script now just calls the bernoulli() builtin
    function.  It remains for backward compatibility.

    The Makefile now builds have_fpos_pos.h to determine if the
    a non-scalar FILEPOS has a __pos stucture element.  If it does,
    the FILEPOS_BITS is taken to be the size of just the __pos element.

    Misc fixes related to non-scalar (e.g., structure) FILEPOS.  Fixed
    a compile problems where non-scalar FILEPOS were incorrectly assigned.

    Fixed make depend rule.

    Return an error on malloc / realloc failures for bernoulli and
    euler functions.

    Added MAKEFILE_REV make variable to help determine Makefile version.
    Fixed the way the env rule reports Makefile values.


The following are the changes from calc version 2.11.3t0 to 2.11.4:

    Increased the maximum number of args for functions from 100 to 1024.
    Increased calc's internal evaluation stack from 1024 to 2048 args.
    Added test8600.cal to the regression suite to test these new limits.

    Updated and fixed misc typos in calc/README.

    Clarified in the COPYING file that ALL calc source files, both
    LGPL covered and exceptions to the LGPL files may be freely used
    and distributed.

    Added help files or updated for: bernoulli, calc_tty, catalan,
    digit, digits, euler, freeeuler, places and sleep.

    A collection of 18 patches from Ernest Bowen
    <ernie at turing dot une dot edu dot au>:

    (1)  A new flag -f has been defined which has the effect of a read
    command without the need to terminate the file name with a semicolon
    or newline.  Thus:

	    calc "read alpha; read beta;"

    may be replaced by:

	    calc -f alpha -f beta

    Quotations marks are recognized in a command like

	    calc -f 'alpha beta'

    in which the name of the file to be read includes a space.

    (2) Flags are interpreted even if they are in a string, as in:

	    calc "-q -i define f(x) = x^2;"

    which has the effect of:

	    calc -q -i "define f(x) = x^2;"

    To achieve this, the use of getopts() in calc.c has been dropped in
    favor of direct reading of the arguments produced by the shell.
    In effect, until a "--" or "-s" or a calc command (recognized
    by not starting with '-') is encountered, the quotation signs in
    command lines like the above example are ignored.  Dropping getopts()
    permits calc to specify completely the syntax rules calc will apply
    to whatever it is given by the shell being used.

    (3) For executable script (also called interpreter) files with first
    line starting with "#!", the starting of options with -S has been
    replaced by ending the options with -f.  For example, the first line:

	    #! full_pathname_for_calc -S -q -i

    is to be replaced by:

	    #! full_pathname_for_calc -q -i -f

    Thus, if the pathname is /usr/bin/calc and myfile contains:

	    #!/usr/bin/calc -q -i -f
	    global deg = pi()/180;
	    define Sin(x) = sin(x * deg);

    and has been made executable by:

	    chmod u+x myfile

    myfile would be like a version of calc that ignored any startup
    files and had an already defined global variable deg and a function
    Sin(x) which will return an approximation to the sine of x degrees.
    The invocation of myfile may be followed by other options (since
    the first line in the script has only flagged options) and/or calc
    commands as in:

	    ./myfile -c read alpha '; define f(x) = Sin(x)^2'

    (The quotation marks avoid shell interpretation of the semicolon and
    parentheses.)

    (4) The old -S syntax for executable scripts implied the -s flag so that
    arguments in an invocation like

	    ./myfile alpha beta

    are passed to calc; in this example argv(0) = 'alpha', argv(1) =
    'beta'.  This has been changed in two ways: an explicit -s is required
    in the first line of the script and then the arguments passed in the
    above example are argv(0) = 'myfile', argv(1) = 'alpha', argv(1) = 'beta'.

    In an ordinary command line, "-s" indicates that the shell words
    after the one in which "-s" occurred are to be passed as arguments
    rather than commands or options.  For example:

	    calc "-q -s A = 27;" alpha beta

    invokes calc with the q-flag set, one command "A = 27;", and two arguments.

    (5) Piping to calc may be followed by calc becoming interactive.
    This should occur if there is no -p flag but -i is specified, e.g.:

	    cat beta | calc -i -f alpha

    which will do essentially the same as:

	    calc -i -f alpha -f beta

    (6) The read and help commands have been  changed so that several
    files may be referred to in succession by separating their names
    by whitespace.  For example:

	    ; read alpha beta gamma;

    does essentially the same as:

	    ; read alpha; read beta; read gamma;

    This is convenient for commands like:

	    calc read file?.cal

    when file?.cal expands to something like file1.cal file2.cal file3.cal:

	    myfiles='alpha beta gamma'
	    calc read $myfiles

    or for C-shell users:

	    set myfiles=(alpha beta gamma)
	    calc read $myfiles


    (7) The -once option for read has been extended to -f.  For example,

	    calc -f -once alpha

    will ignore alpha if alpha has been read in the startup files.  In a
    multiple read statement, -once applies only to the next named file.
    For example

	    ; read -once alpha beta -once gamma;

    will read alpha and gamma only if they have not already been read,
    but in any case, will read beta.

    (8) A fault in the programming for the cd command has been corrected
    so that specifying a directory by a string constant will work.  E.g:

	    ; cd "my work"

    should work if the current directory has a directory with name "my work".

    (9) new functions bernoulli(n) and euler(n) have been defined to
    return the Bernoulli number and the Euler number with index n.
    After evaluation for an even positive n, this value and these for
    smaller positive even n are stored in a table from which the values
    can be reread when required.  The memory used for the stored values
    can be freed by calling the function freebernoulli() or freeeuler().

    The function catalan(n) returns the catalan number with index n.
    This is evaluated using essentially comb(2*n, n)/(n+1).

    (10) A function sleep(n) has been defined which for positive n calls
    the system function sleep(n) if n is an integer, usleep(n) for other
    real n.  This suspends operation for n seconds and returns the null
    value except when n is integral and the sleep is interrupted by a
    SIGINT, in which case the remaining number of seconds is returned.

    (11) The effect of config("trace", 8) which displays opcodes of
    functions as they are successfully defined has been restricted to
    functions defined with explicit use of "define".  Thus, it has been
    deactivated for the ephemeral functions used for evaluation of calc
    command lines or eval() functions.

    (12) The functions digit(), digits(), places() have been extended to
    admit an optional additional argument for an integral greater-than-one
    base which defaults to 10.  There is now no builtin limit on the
    size of n in digit(x, n, b), for example, digit(1/7, -1e100) which
    would not work before can now be handled.

    (13) The function, digits(x), which returns the number of decimal
    digits in the integer part of x has been changed so that if abs(x) <
    1, it returns 0 rather than 1.  This also now applies to digits(x,b).

    (14) Some programming in value.c has been improved.  In particular,
    several occurrences of:

	    vres->v_type = v1->v_type;
	    ...
	    if (v1->v_type < 0) {
		    copyvalue(v1, vres);
		    return;
	    }

    have been replaced by code that achieves exactly the same result:

	    vres->v_type = v1->v_type;
	    ...
	    if (v1->v_type < 0)
		    return;

    (15) Some operations and functions involving null-valued arguments
    have been changed so that they return null-value rather than "bad
    argument-type" error-value.  E.g. null() << 2 is now null-valued
    rather than a "bad argument for <<" error-value.

    (16) "global" and "local" may now be used in expressions.  For example:

	    ; for (local i = 0; i < 5; i++) print i^2;

    is now acceptable, as is:

	    ; define f(x = global x) = (global x = x)^2;

    which breaks wise programming rules and would probably better be handled
    by something like:

	    ; global x
	    ; define f(t = x) = (x = t)^2;

    Both definitions produce the same code for f.  For non-null t, f(t)
    returns t^2 and assigns the value of t to x;  f() and f(t) with null t
    return x^2.

    Within expressions, "global" and "local" are to be followed by just one
    identifier.  In "(global a = 2, b)" the comma is a comma-operator; the
    global variable a is created if necessary and assigned the value 2, the
    variable b has to already exist.   The statement "global a = 2, b" is
    a declaration of global variables and creates both a and b if they
    don't already exist.

    (18) In a config object, several components have been changed from
    long to LEN so that they will now be 32 bit integers for machines with
    either 32 or 64-bit longs.  In setting such components, the arguments
    are now to less than 2^31.  Before this change:

	    ; config("mul2", 2^32 + 3)

    would be accepted on a 64-bit machine but result in the same as:

	    ; config("mul2", 3)


The following are the changes from calc version 2.11.2t0 to 2.11.2t1.0:

    Fixed a bug whereby help files are not displayed correctly on
    systems such as NetBSD 1.4.1.  Thanks to a fix from Jakob Naumann.

    Changed EMail addresses to use asthe.com.  Changed URLs to use
    www.isthe.com.  NOTE: The EMail address uses 'asthe' and the web
    site URL uses 'isthe'.

    Using calc-bugs at asthe dot com for calc bug reports,
    calc-contrib at asthe dot com for calc contributions,
    calc-tester-request at asthe dot com for requests to join calc-tester and
    calc-tester at asthe dot com for the calc tester mailing list.

    Replaced explicit EMail addresses found this file with the <user at
    site dot domain> notation to reduce the potential for those folks
    to be spammed.

    The Makefile attempts to detect the existence of /dev/urandom with -e
    instead of the less portable -c.

    Misc Makefile fixes.


The following are the changes from calc version 2.11.1t3 to 2.11.1t4:

    Removed non-portable strerror() tests (3715, 3724 and 3728) from
    calc/regress.cal.

    Fixed missing strdup() from func.c problem.

    Fixed a problem that would have come up on a very long #! command line
    if the system permitted it.


The following are the changes from calc version 2.11.1 to 2.11.1t2.2:

    Placed calc under version 2.1 of the GNU Lesser General Public License.

	The calc commands:

	    help copyright
	    help copying
	    help copying-lgpl

	should display the generic calc copyright as well as the contents
	of the COPYING and COPYING-LGPL files.

	Those files contain information about the calc's GNU Lesser General
	Public License, and in particular the conditions under which you
	are allowed to change it and/or distribute copies of it.

    Removed the lint facility from the Makefile.  Eliminated Makefile
    variables: ${LCFLAGS}, ${LINT}, ${LINTLIB} and ${LINTFLAGS}.
    Removed the lint.sed file.

    Cleaned up help display system.  Help file lines that begin with
    '##' are not displayed.

    Calc source and documentation now uses the the these terms:

	*.cal files	calc resource file
	*.a files	calc binary link library
	#! files	calc shell script

    Renamed 'help stdlib' to 'help resource'.	The 'help stdlib' is
    aliased to 'help resource' for arg compatibility.

    Renamed config("lib_debug") to config("resource_debug").
    The config("lib_debug") will have the same effect as
    config("resource_debug") for backward compatibility.

    Renamed the source sub-directory lib to cal.  The default $CALCPATH
    now uses ./cal:~/cal (instead of ./lib:~/lib).  Changed LIB_PASSDOWN
    Makefile variable to CAL_PASSDOWN.

    Fixed misc compile warnings and bugs.

    Fixed problem of incorrect paths in the formation of installed
    calc shell scripts.

    Changed the recommended Comqaq cc compile to be -std0 -fast -O4 -static.

    Fixed a problem related to asking for help for a non-existent file.

    Added ./.calcinit to the default calcrc.

    Added cscript/README and help cscript to document the calc shell
    script supplied with calc.


The following are the changes from calc version 2.11.0t10 to 2.11.0t11:

    Misc code cleanup.	Removed dead code.  Removed trailing whitespace.
    Fixed whitespace to make the best use of 8 character tabs.

    Fixed some bugs relating to '// and %' in combination with some
    of the the rounding modes based on a patch from Ernest Bowen
    <ernie at turing dot une dot edu dot au>.

    A patch from Klaus Alexander Seistrup <klaus at seistrup dot dk>, when
    used in combination with the GNU-readline facility, will prevent
    it from saving empty lines.

    Minor typos fixed in regress.cal

    Added 8500 test series and test8500.cal to perform more extensive
    tests on // and % with various rounding modes.

    The 'unused value ignored' messages now start with Line 999: instead
    of just 999:.

    Fixed the long standing issue first reported by Saber-C in the
    domul() function in zmil.c thanks to a patch by Ernest Bowen
    <ernie at turing dot une dot edu dot au>.

    Added zero dimensional matrices.  A zero dimensional matrix is defined as:

	mat A[]	  or	A = mat[]

    Updated the help/mat file to reflect the current status of matrices
    including zero dimensional matrices.

    Added indices() builtin function as written by Ernest Bowen <ernie
    at turing dot une dot edu dot au> developed from an idea of Klaus
    Seistrup <klaus at seistrup dot dk>.  See help/indices for details.

    Fixed a number of insure warnings as reported by Michel van der List
    <vanderlistmj at sbphrd dot com>.

    Fixed a number of help file typos discovered by Klaus Alexander
    Seistrup <klaus at seistrup dot .dk>.

    Removed REGRESS_CAL as a Makefile variable.

    Added calcliblist and calcliblistfmt utility Makefile rules to allow
    one to print the list of distribution files that are used (but not
    built) to form either the libcalc.a or the libcustcalc.a library.

    Added a patch from <Randall.Gray at marine dot csiro dot au> to make
    ^D terminate, but *only* if the line it is on is completely empty.
    Removed lib/altbind and removed the CALCBINDINGS Makefile variable.

    A new config("ctrl_d") value controls how the ``delete_char'', which
    by default is bound to ^D (Control D), will or will not exit calc:

	config("ctrl_d", "virgin_eof")

	    If ^D is the only character that has been typed on a line,
	    then calc will exit.  Otherwise ^D will act according to the
	    calc binding, which by default is a Emacs-style delete-char.

	    This is the default mode.

	config("ctrl_d", "never_eof")

	    The ^D never exits calc and only acts according calc binding,
	    which by default is a Emacs-style delete-char.

	    Emacs purists may want to set this in their ~/.calcrc startup file.

	config("ctrl_d", "empty_eof")

	    The ^D always exits calc if typed on an empty line.	 This
	    condition occurs when ^D either the first character typed,
	    or when all other characters on the line have been removed
	    (say by deleting them).

	    Users who always want to exit when ^D is typed at the beginning
	    of a line may want to set this in their ~/.calcrc startup file.

	Note that config("ctrl_d") apples to the character bound to each
	and every ``delete_char''.  So if an alternate binding it setup,
	then those char(s) will have this functionality.

    Updated help/config and help/mode, improved the readability and
    fixed a few typos.	Documented modes, block formats and block bases
    ("mode", "blkfmt" & "blkbase") that were previously left off out of
    the documentation.

    The config("blkbase") and config("blkfmt") values return strings
    instead of returning integers.  One cannot use integers to set
    these values, so returning integers was useless.

    Applied the dangling name fix from Ernest Bowen
    <ernie at turing dot une dot edu dot au>.

    Show func prints function on order of their indices, and with
    config("lib_debug") & 4 == 4  some more details about the functions
    are displayed.

    Fixed another ``dangling name'' bug for when the object types list
    exceeded 2000.

    Fixed a bug related to opening to a calc session:

	define res_add(a,b) = obj res {r} = {a.r + b.r};
	...
	obj res A = {1,2}. obj res B = {3,4}

    A hash of an object takes into account the object type.  If X and Y
    are different kinds of objects but have the same component values,
    they will probably return different rather than the same values for
    hash(X) and hash(Y).

    Added support for config("ctrl_d") to the GNU-readline interface
    as written by Klaus Alexander Seistrup <klaus at seistrup dot dk>.

	Currently, the config("ctrl_d", "virgin_eof") is not fully
	supported.  Under GNU-readline, it acts the same way as
	config("ctrl_d", "empty_eof").	Emacs users may find this
	objectionable as ``hi^A^D^D^D'' will cause calc to exit due to
	the issuing of one too many ^D's.

	Emacs users may want to put:

	    config("ctrl_d", "never_eof"),;

	into their ~/.calcrc startup files to avoid this problem.

    Made misc documentation fixes.

    Fixed the make depend rule.

    Applied Ernest Bowen's <ernie at turing dot une dot edu dot au>
    complex function power(), exp() and transcendental function patch:

	Calc will return a "too-large argument" error-value for exp(x,
	epsilon) if re(x) >= 2^30 or if an estimate indicates that the
	result will have absolute value greater than 2^2^30 * epsilon.
	Otherwise the evaluation will be attempted but may fail due to
	shortage of memory or may require a long runtime if the result
	will be very large.

	The power(a, b, epsilon) builtin will return a "too-large result"
	if an estimate indicates that the result will have absolute value
	that is > 2^2^30 * epsilon.  Otherwise the evaluation will be
	attempted but may fail due to shortage of memory or may require
	a long runtime if the result will be very large.

	Changes have been made to the algorithms used for some special
	functions sinh(), cosh(), tanh(), sin(), cos(), etc., that make
	use of exp().  In particular  tanh(x)  is now much faster and
	doesn't run out of memory when x is very large - the value to
	be returned is then 1 to a high degree of accuracy.

	When the true value of a transcendental function is 1, as is
	cos(x) for x == 0, calc's version of the function will now return
	1 rather than the nearest multiple of epsilon.	E.g. cos(0, 3/8)
	no longer returns 9/8.

	The restriction of abs(n) < 1000000 on scale(x, n) has been
	removed.  The only condition n now has to satisfy for calc to
	attempt the operation is  n < 2^31, the same as for calc to
	attempt x << n and x^n.

	Changed root(x,n) so that when x is negative and n is odd it
	returns the principal complex n-th root of x rather than -1, e.g.
	root(-1,3) now returns -.5+.8660...i.

	Changed power(a,b) to permit a to be negative when b is real.
	E.g. power(-2,3) will now return 8 rather than cause a "negative
	base" error.

	Fixed several improper free and link problems in the comfunc.c code.

    Removed BOOL_B64 symbol from Makefile.

    The following config values return "true" or "false" strings:

	    tilde  tab	leadzero  fullzero  blkverbose	verbose_quit

	These config values can still be set with same boolean strings
	("on", "off", "true", "false", "t", ...) as well as via the
	numerical values 0 (for "false") and non-0 (for "true"), however.

    Added -s to the calc command line.	The -s flag will cause unused
    args (args after all of the -options on the command line) to remain
    as unevaluated strings.

    If calc is called with -s, then the new function argv() will return
    the number of strings on the command line.	Also argv(n) will return
    the n-th such string or null is no such string exists.

    Calc now handles calc shell scripts.  A calc shell script is an
    executable file that starts with:

	    #!/usr/local/bin/calc -S

	Where ``/usr/local/bin/calc'' is the path to the calc binary.
	Additional -options may be added to the line, but it MUST
	start with -S.	For example, the the executable file ``plus''
	contain the following:

	    #!/usr/local/bin/calc -S -e
	    /*
	     * This is a simple calc shell script to add two values
	     */
	    print eval(argv(0)) + eval(argv(1));

	then the following command:

	    ./plus 23 'pi(1e-5)'

	will print:

	    26.14159

    If calc is called with -S as the first arg, then calc will assume that
    it is being called from a #! calc shell script file.  The -S implies
    the -s flag.  If -i is not given, -S also implies -d and -p.

    Fixed the problem with non-literal string type checking for the
    C printf-like functions.  Able to determine if "%ld" or "%lld"
    is appropriate for printing of 64 bit long longs by way of the C
    symbol L64_FORMAT in the longlong.h header file.

    The following lines are treated as comments by calc:

	#! this is a comment
	# this is a comment
	#	this is a comment
	#
	# The lone # above was also a comment
	## is also a comment

    Improved how calc makes changes to file descriptor interactive state.
    Moved state changing code to calc_tty() and orig_tty() in lib_calc.c.
    The libcalc_call_me_last() function will restore all changed descriptor
    states that have not already been restored.

    Added the following read-only config values:

	config("program")	path to calc program or calc shell script
	config("basename")	basename of config("program")
	config("version")	calc version string


The following are the changes from calc version 2.11.0t8.9.1 to 2.11.0t9.4.5:

    The config("verbose_quit") will control the printing of the message:

	    Quit or abort executed

	when a non-interactive ABORT, QUIT or EXIT is encountered.  By default,
	config("verbose_quit") is TRUE and the message is printed.  If one does:

	    config("verbose_quit", 0)

	the message is disabled.

    Added 8400 regression test set and test8400.cal to test the new
    quit and config("verbose_quit") functionality.

    Fixed the BigEndian BASEB==16 regression bugs by correctly swapping
    16 bit HALFs in a 64 bit value (such as a 64 bit file pointer).

    Added calclevel() builtin to calculation level at which it is called.

    Added help/calclevel and help/inputlevel help files.

    Removed regression tests 951 and 5984 so that the regress test will
    run in non-interactively / without a TTY such as under Debian's
    build daemon.

    The eval(str) builtin will return an error-value rather than cause
    an execution error str has a scan-error.

    Declarations are permitted to end with EOF as well as a newline or ';'.

    When prompt() occurs while reading a file, it will take input from
    the terminal rather than taking it from a file.  For example:

	    /* This demonstrates the use of prompt() and some other things  */
	    config("verbose_quit", 0);
	    define getnumber() {
		local x;
		for (;;) {
		    x = eval(prompt(">>> "));
		    if (isnum(x))
			return x;
		    print "Not a number! Try again";
		}
	    }
	    print "This will display the sqrt of each number you enter";
	    print "Enter quit to stop";
	    for (;;) {
		print sqrt(getnumber());
	    }
	    print "Good bye";

	Comments entered at input terminal level may be spread over several
	lines.	For example:

	    /*
	     * Assume that this calc script is called: comment.cal
	     * Then these commands now work:
	     *	cat comment.cal | calc
	     *	calc < comment.cal
	     */
	    print "Hello";

    Added:

	-D calc_debug[:lib_debug:[user_debug]]

    to set the initial value of config("calc_debug"), config("lib_debug")
    and config("user_debug").

    The : separated strings of -D are interpreted as signed 32 bit values.
    After an optional leading sign a leading zero indicates octal
    conversion, and a leading ``0x'' or ``0X'' hexadecimal conversion.
    Otherwise, decimal conversion is assumed.

    Reordered the config structure moving calc_debug ahead of lib_debug.

    Added bits 4 and 5 to config("calc_debug"):

	4	Report on changes to the state of stdin as well as changes
		to internal variables that control the setting and restoring
		of stdin.

	5	Report on changes to the run state of calc.

    Fixed portability issue in seed.c relating to /dev/urandom and ustat.

    Added a fix from Martin Buck <mb at netwings dot ch> to detect when
    calc aborts early instead of completing the regression test.
    Now 'make chk' will require the last line of calc output to
    end in the string ``Ending regression tests''.

    Added a patch from Martin Buck <mb at netwings dot ch> to allow use of
    GNU-readline.  Note that GNU-readline is not shipped with calc.
    His patch only provides the hooks to use it.  One must comment out:

	    USE_READLINE=
	    READLINE_LIB=
	    READLINE_INCLUDE=

	and comment in:

	    USE_READLINE= -DUSE_READLINE
	    READLINE_LIB= -lreadline -lhistory
	    READLINE_INCLUDE= -I/usr/include/readline

	in addition to pre-installing GNU-readline in your system to use
	this facility.

    Changed the "object already defined" math_error to a scanerror message.

    Removed the limit on the number of object types.

    Calc tarballs are now named calc-version.tar.gz and untar into
    a sub-directory called calc-version.

    Made a small change to declarations of static variables to reduce
    the internal opcodes needed to declare them.

    Fixed a permission problem on ranlib-ed *.a files that was reported
    by Michael Somos <somos at grail dot cba dot csuohio dot edu>.

    Added patch by Klaus Alexander Seistrup <klaus at seistrup dot dk>
    related to GNU-readline:

	+ enable calc specific bindings in ~/.inputrc
	+ save a copy of your session to disk and reload them next
	  time you're using calc
	+ only add a line to the history if it is different from
	  the previous line

    Added the Makefile symbol HAVE_GETRUSAGE to determine if the
    system supports the getrusage() system call.

    Fixed the make depend code in the custom and sample Makefiles.

    Fixed how the help/builtin file is formed.	The help/Makefile is
    now given the name of the native C compiler by the top level Makefile.

    The include files are installed under INCDIRCALC (a new Makefile variable)
    which by default is ${INCDIR}/calc.	 The INCDIR (also a new Makefile var)
    by default is /usr/local/include.  Include files previously installed
    directly under ${LIBDIR} will be removed.

    Added the piforever() function to lib/pi.cal.  It was written by
    Klaus Alexander Seistrup <klaus at seistrup dot dk> and was inspired by
    an algorithm conceived by Lambert Meertens.	 (See also the ABC
    Programmer's Handbook, by Geurts, Meertens & Pemberton, published
    by Prentice-Hall (UK) Ltd., 1990.)	The piforever() function prints
    digits of pi for as long as your memory and system uptime allows.  :-)

    Fixed the URLs found thruout the source and documentation which did
    not and in /, but should for performance and server load reasons.

    Cleaned up and improved handling of "mat" and "obj".  The comma in:

	    mat A[2], B[3];

	is changed to whatever is appropriate in the context:

	    + comma operator
	    + separator of arguments in a function call
	    + separator of arguments in a definition
	    etc.

	The expression (mat A[2]), B[3] returns B[3], assuming B already
	exists as something created by a statement like: global mat B[4].

	What used to be done by the expression:

	    mat A[2], B[3]

	will now require something like:

	    mat A[2], mat B[3]	or	A = mat[2], B = mat[3]

	For example, if obj point and obj pair are known types, the
	following is now allowed:

	    L = list(mat[2], mat[3], obj point, obj pair)

	As another example, the following is allowed:

	    define f(a = mat[2] = {3,4}) = 5 * a;

	as well as the following:

	    obj point {x,y}, PP = obj pair {A,B} = {obj point, obj point}

	which creates two object types at compile time and when executed,
	assigns a pair-object value to a variable PP.

    Fixed a bug whereby a for loop would behave incorrectly.  For example:

	    config("trace", 2),
	    global x;
	    define f() {for ( ; x > 0; x--) {print x;}}
	    x = 5, f()

	will stop after printing 1 instead of looping forever.

    Added values l_format, which when CHECK_L_FORMAT is defined ahead
    of including longlong.h will help detect when a system can deal with
    'long long' but not '%lld' in printf.  If a system with 'long long'
    uses '%ld' to print a 64 bit value, then l_format will be > 0;
    otherwise if "%lld" is required, l_format will be < 0.

    Added HAVE_STRDUP Makefile variable as well as the have_strdup.c
    program that forms the have_strdup.h file.	The have_strdup.h file
    will define HAVE_STRDUP is the system has strdup().	 If HAVE_STRDUP
    is not defined, then calc will use calc_strdup() to simulate
    the real strdup() function.

    Calc no longer makes use of sys_errlist and sys_nerr.  Some systems
    no longer support these values (even though they should from a
    legacy prospective).  Calc now relies on the fact that strerror()
    will return NULL of no such system error exists.  System errors >=
    10000 will be considered calc errors instead.  The Makefile symbol
    ERRNO_DECL has gone away as well as calc_errno.c and calc_errno.h.

    System errors that are are not known to to the libc strerror()
    function, will now print (via the strerror() calc builtin function)
    something such as:

	    Unknown error 9999

    Fixed some insure code inspection tool issues that were discovered
    and investigated by Michel van der List <vanderlistmj at sbphrd dot com>.

    Made an effort to ensure that the v_subtype of VALUES are initialized
    to V_NOSUBTYPE thruout the source code.

    Established a separate calc-bugs address from the calc-tester
    mailing list.  Using anti-spam address forms in order to try and
    stay under the radar of spammers as much as one can do so.


The following are the changes from calc version 2.11.0t8 to 2.11.0t8.9:

    Moved 'wishlist' enhancements from the help/todo file to a new
    help/wishlist file.	 Ordered, by priority, help/todo items into
    Very High, High and Medium priority items.

    The BUGS file now has a 'bugs' section as well as a 'mis-features'
    section.

    Improved how calc internally dealt with reading EOF or '\0' characters.

    Calc now allows multiple defines to occur on the same line:
    (Thanks goes to Ernest Bowen <ernie at turing dot une dot edu dot au>)

	define f8300(x) = x^2; define g8300(x) = 1 - x;

    Improved calc's ability to deal with and recover from errors.

    Added inputlevel() builtin to return the input processing level.
    In an interact mode, inputlevel() returns 0.  When directly reading
    a calc script, inputlevel() returns 1.  When reading a script which
    in turn reads another script, inputlevel() returns 2.  etc...

    If $CALCRC has more than one file as in file1:file2 and an error
    occurs in file1, then calc -c will not read file2.

    Fixed some of the old EMail addresses found in calc documentation.

    Added HAVE_USTAT, HAVE_GETSID, HAVE_GETPGID, HAVE_GETTIME, HAVE_GETPRID
    and HAVE_URANDOM symbols to the Makefile.  These symbols, along with
    have_ustat.c, have_getsid.c, have_getpgid.c, have_gettime.c and
    have_getprid.c form: have_ustat.h, have_getsid.h, have_getpgid.h,
    have_gettime.h, have_getprid.h and have_urandom.h which in turn
    are used by pseudo_seed() in seed.c to determine what types of
    system services can be used to form a pseudo-random seed.

    Fixed the way calc -c will continue processing $CALCRC when errors
    are encountered.  Unless -d is also given, calc -c will report
    when calc is unable to open a $CALCRC file.

    Fixed the lower level make depend rules.

    Misc cleanup on the have_*.c support source files.

    Misc source file cleanup for things such as } else { style consistency.

    Fixed the basis for FNV-1 hashes.  Prior to this fix, the hash()
    builtin produced FNV hash values that did not match the FNV-1
    algorithm as specified in:

	http://www.isthe.com/chongo/tech/comp/fnv/index.html

    Removed an unused argument in the function getbody() in codegen.c.

    Encountering of EOF in getbody() will cause a scanerror rather then
    stop activity.  This will now result in a scanerror:

	echo 'define f(x) { ' > myfile
	calc -i read myfile

    A '{' at the start of a command and a later matching '}' surrounding zero
    or more statements (and possibly newlines) results in a function body to
    be "evaluated".   This permits another command to follow on the same
    line as the '}' as in:

		{display(5)} read something;
	and:
		{static a = 5} define f(x) = a + x;

    String constants can now be concatenated.  For example:

	s = "curds" ' and ' "whey";

    Added FNV hash to the regression test suite.

    Added Ernest Bowen's <ernie at turing dot une dot edu dot au> fix for the
    FNV regression test of the hash() builtin function.

    Added Ernest Bowen's <ernie at turing dot une dot edu dot au> patch to
    improve the way config("calc_debug").  Now the lower 4 bits of the
    config("calc_debug") parameter have the following meaning:

       n	       Meaning of bit n of config("calc_debug")

       0       Outputs shell commands prior to execution.

       1       Outputs currently active functions when a quit instruction
	       is executed.

       2       Some details of shs, shs1 and md5 hash states are included
	       in the output when these are printed.

       3       When a function constructs a block value, tests are
	       made that the result has the properties required for use of
	       that block, e.g. that the pointer to the start of the
	       block is not NULL, and that its "length" is not negative.
	       A failure will result in a runtime error.

    Changed the meaning of (config("calc_debug") & 1) from only printing
    the shell commands (and pausing) while displaying help files into
    the printing of any shell command prior to execution.

    Documented the meaning of config("lib_debug"):

	n		Meaning of bit n of config("lib_debug")

	0	When a function is defined, redefined or undefined at
		interactive level, a message saying what has been done
		is displayed.

	1	When a function is defined, redefined or undefined during
		the reading of a file, a message saying what has been done
		is displayed.

	The value for config("lib_debug") in both oldstd and newstd is
	3, but if calc is invoked with the -d flag, its initial value
	is zero.  Thus, if calc is started without the -d flag, until
	config("lib_debug") is changed, a message will be output when a
	function is defined either interactively or during the reading
	of a file.

    Changed the calc lib files to reflect the new config("lib_debug")
    bit field meaning.	Calc lib files that need to print extra information
    should now do something such as:

	if (config("lib_debug") & 3) {
		print "obj xyz defined";
		print "funcA([val1 [, val2]]) defined";
		print "funcB(size, mass, ...) defined";
	}

    Fixed the help/custom_cal, help/new_custom, and help/copy files so
    that they contain the correct contents instead of the 'usage' file.

    Fixed problem with loss of bindings when calc -i args runs into
    an error while processing 'args' and drops into interactive mode
    without the terminal bindings being set.

    Added patch from Ernest Bowen to establish the abort command as
    well as to clarify the roles of quit and exit.  See the help/command
    file for details.

    Updated to some extend, the help/statement and help/command help
    files with new information about SHOW, QUIT, EXIT and ABORT.

    Added show sizes to pzasusb8.cal.

    Updated calc man page and help/usage file to reflect recent
    command line changes.

    Fixed a bug, reported by Michael Somos <somos at grail dot cba dot
    csuohio dot edu>, which prevented calc -m from being used.

    Fixed misc compiler warnings.


The following are the changes from calc version 2.11.0t7 to 2.11.0t7.5:

    Calc has some new command line flags / command line meaning:
    (Thanks goes to Ernest Bowen <ernie at turing dot une dot edu dot au>)

	-i	Go into interactive mode if possible.

	-c	Continue reading command lines even after an execution
		error has caused the abandonment of a line

	To understand the -i and -c effects, consider the following
	file (call it myfile.cal) which has deliberate errors in it:

	    print 1;
	    mat A[1] = {2,3};
	    print 2;
	    epsilon(-1);
	    print 3;

	calc read myfile

	    Reports an error on the 2nd line and exits; prints 1 only.

	calc -c read myfile

	    Report errors on the 2nd and 4th lines and exits; prints 1,2 and 3.

	calc -i read myfile

	    Report errors on the 2nd and gives you a prompt; prints 1 only.

	calc -i -c read myfile

	    Report errors on the 2nd and 4th and gives you a prompt;
	    prints 1, 2 and 3.

	cat myfile | calc

	    Reports an error on the 2nd line and exits; prints 1 only.

	cat myfile | calc -c

	    Report errors on the 2nd and 4th lines and exits; prints 1,2 and 3.

	Note that continuation refers to command lines, not to statements.  So:

	    calc -c 'print "start"; mat A[1] = {2,3}; print "end";'

	since it contains no newline, the whole string is compiled,
	but execution is abandoned when the error is encountered and
	the string ``end'' is not printed.

	You can use your shell to supply newlines in your command line
	arguments.  For example in sh, ksh, bash:

	    calc -c 'print "start";
	    mat A[1] = {2,3};
	    print "end";'

	will print both ``start'' and ``end''.	C-shell users can do:

	    calc -c 'print "start"; \
	    mat A[1] = {2,3}; \
	    print "end";'

	however sh, ksh, bash will not see ``end'' printed because their
	shell will remove the internal newlines.

    Added display(n) builtin which does almost the same as config("display",n)
    except that rather than causing an execution with an out-of-range or
    bad-type argument type, it simply writes a message to stderr.  This
    also now happens to the errmax() builtin.

    Added qtime.cal to the standard calc library.

    Added another command line flag to calc:

	-d	Disable display of the opening title and config("lib_debug",0)

	The command:

	    calc 'read qtime; qtime(2)'

	will output something like:

	    qtime(utc_hr_offset) defined
	    It's nearly ten past six.

	whereas:

	    calc -d 'read qtime; qtime(2)'

	will just say:

	    It's nearly ten past six.

    A call of errmax(-1) will prevent errcount from aborting calc.

    Add the function stoponerror(n) which, as the name implies, controls
    if calc stop on an error based on the value of n:

	n > 0	stop on error even if -c was given on the command line
	n == 0	if -c, continue, without -c, stop
	n < 0	continue on error, even if -c was given on the command line

    Calc compilation now stops at the first scanerror.

    Restored the feature where -p disables the printing of leading tabs
    as of config("tab",0) had been executed.  So using calc in a pipe:

	calc -p 2+17 | whey

    will write '19' instead of '\t19' to the whey command.

    Updated calc man page and help/usage file to reflect recent
    command line changes.

    Converted start_done into a general calc run state enum called
    run_state within the calc source.

    Removed README.OLD.

    Added the Makefile variable ${LCC} to invoke the local c compiler.
    By default, ${CC} also run the ${LCC} compiler.  The distinction is
    useful when using something such as purify.	 In the case of ${LCC},
    only the local C compiler is invoked.  In the case of ${CC} a purify
    compile is invoked.	 Only the source that must be compiled and run
    on the local machine use ${LCC}; everything else uses ${CC}.

    Fixed memory buffer related problem in eatstring() in token.c.

    Fixed memory leaks related to putenv().

    Fixed memory leaks related to srandom().

    Fixed compilation warnings and problems on BSDI.

    Removed ${CCMAIN} as a variable from the Makefile.	Now files
    use either ${CFLAGS} for general C source and ${ICFLAGS} for
    intermediate C source (e.g., special code for building hsrc files).

    The main calc URL is now:

	http://www.isthe.com/chongo/tech/comp/calc/

    Misc calc man page fixes.


The following are the changes from calc version 2.11.0t1 to 2.11.0t6.3:

    Removed the makefile symbol MAIN.  Now forcing all functions to correctly
    be declared main.  To satisfy some old broken compilers, a return 0;
    (instead of an exit(0);) is used at the end of main().

    A few of files that were added to calc used 4 character indentation
    whereas most of calc uses 8 character indentation.	These imported
    sources have been changed to conform better with the calc style.

    Added the program calc_errno.c and the Makefile symbol ERRNO_DECL.
    If ERRNO_DECL is empty, calc_errno.c will try various ways to
    declare errno, sys_errlist and sys_nerr.  On success or when
    it gives up, calc_errno will output the middle of the calc_errno.h
    header file.  If ERRNO_DECL is  -DERRNO_NO_DECL, or -DERRNO_STD_DECL
    or -DERRNO_OLD_DECL then the Makefile will build the middle
    of the calc_errno.h header file without calc_errno.c's help.

    The func.c file now includes the constructed header file calc_errno.h
    to ensure that errno, sys_errlist and sys_nerr are declared correctly.

    Changed check.awk to be more 'old awk' friendly.

    Made some of the source a little more ++ friendly.	We are NOT
    porting calc to C++!  We will NOT support C++ compilation of calc.
    Calc will written ANSI C.  We just compiled with a suggestion from
    Love-Jensen, John <jlove-jensen at globalmt dot com> to make calc's version
    of C a little more to C++ compilers.  We are simply avoiding symbols
    such as new or try for example.

    Renamed README to README.OLD.  Renamed README.FIRST to README.
    Updated README, lib/README and BUGS to reflect new URLs and addresses.

    Added a HOWTO.INSTALL file.

    Reordered cc Makefile variable sets in the main Makefile.

    Fixed a bug in hnrmod() and applied a fix that was reported by Ernest
    Bowen <ernie at turing dot une dot edu dot au>.  Added regression tests
    1103 to 1112 to confirm the fix.

    Fixed a bug in version.c related to MINOR_PATCHs in both the
    empty and non-empty MINOR_PATCH cases.

    Fixed malloc and bad storage issues reported by Michel van der List
    <vanderlistmj at sbphrd dot com>.

    Fixed some problems related to path processing while opening files.
    Under extreme cases, an excessively long filename or CALCPATH value
    could create problems.  Placed guards in opensearchfile() function
    in input.c to catch these cases.

    Fixed cases were malloc failures were silently ignored in input.c.

    Eliminated the PATHSIZE limit and the PATHSIZE symbol.

    Added MAX_CALCRC to limit the length of the $CALCRC environment
    variable to 1024 chars.

    Fixed the magic number relating to the initial number of constants
    declared by initconstants().  It is now related to the length
    of the initnumbs[] NUMBER array.

    Added a 'Dec Alpha / Compaq Tru64 cc (non-gnu) compiler set'
    section to the main Makefile.

    Fixed a string handling bug discovered by Dr.D.J.Picton
    <dave at aps5 dot ph dot bham dot ac dot uk> in the custom demo code.

    Fixed a bug in the hnrmod() builtin that was discovered by
    Ernest Bowen <ernie at turing dot une dot edu dot au>.

    Added FORCE_STDC symbol.  When defined it will force __STDC__ like
    conditions.	 Thus for compilers with as the Solaris cc compiler
    that are ANSI-like but still define __STDC__ as 0, one can use
    -DFORCE_STDC and make use of ANSI-like features.

    Removed the CCSHS symbol from the Makefile.	 The shs.c and shs1.c
    files are now compiled with full ${CFLAGS}.

    The custom.c file is now compiled with full ${CFLAGS}.

    Rewrote command line / argument processing code.  Calc is now
    using getopt(3) argument processing.

    Fixed a memory leak related to converting strings to numbers
    in the str2q() function in qio.c.

    Fixed a problem with reading uninitialized memory in the
    v_subtype of a VALUE in the copyvalue() function in value.c.

    Fixed problems in func.c where temporary VALUEs were not
    having their v_type elements initialized.

    Fixed a memory leak in qpi() in qtrans.c.

    Fixed a memory leak in math_getdivertedio() in zio.c.

    Fixed a problem with points going beyond the end of allocated
    memory in addstring() in string.c.

    Fixed a memory leak in zgcdrem(), f_putenv(), zlog() and
    zlog10() in zfunc.c.

    Fixed a memory leak in zdiv() and zshift() in zmath.c.

    Fixed memory leaks in zsrand() in zrand.c.

    Fixed a memory leak in zsrandom1() in zrandom.c.  Fixed memory
    leaks associated with replacing the internal random state with
    another random state.

    Added seed() builtin to return a 64 bit seed for a
    pseudo-random generator.

    Added functionality from Ernest Bowen <ernie at turing dot une dot
    edu dot au> to permit nested "= {...}" assignments for lists as well
    as matrices and objects.  Now one can have a list, matrix or object,
    some of whose elements are lists, matrices or objects, to any depth
    of recursion, and assign values to any number of particular elements
    by an appropriate "initialization" expression.  For example:

	A = mat[2] = {list(1,2), list(3,4,list(5,6))};

    and then assign values to the 6 number elements by:

	A = {{7,8}, {9,10,{11,12}}};

    Closed files that were previously left open from test4600.cal
    as executed by regress.cal and from opening /dev/null by
    regress.cal itself.

    Fixed memory leaks from f_strprintf() and f_putenv() in func.c.

    The regress.cal test suite calls freeredc(), freestatics() and
    freeglobals() at the end of the test suite to free storage
    consumed during the regression.

    Added custom function custom("pzasusb8", n) and lib/pzasusb8.cal based on
    Ernest Bowen's diagnostic patch.

    Thanks to the efforts of Ernest Bowen <ernie at turing dot une dot
    edu dot au> and Dr.D.J.Picton <dave at aps5 dot ph dot bham dot ac
    dot uk>, a nasty endian-ness bug in the sha and sha1 hash functions
    that showed up on machines such as the Sparc was fixed.

    Added functionality from Ernest Bowen <ernie at turing dot une
    dot edu dot au> to give arguments as well as function names after
    definitions when config("lib_debug") >= 0.

    Removed if (config("lib_debug") >= 0) { ... } the ends of most
    of the calc library scripts because it was redundant with the
    new config("lib_debug") >= 0 functionality.	 Some of the calc
    library still has a partial section because some useful
    additional information was being printed:

	chrem.cal	deg.cal	     lucas_tbl.cal   randrun.cal
	mfactor.cal	mod.cal	     poly.cal	     seedrandom.cal
	surd.cal	varargs.cal

    Fixed ellip.cal so that its defined function does not conflict with
    the factor() builtin function.

    Fixed mod.cal so that a defined function does not conflict with
    the mod() builtin function.

    The regression test suite now reads in most calc libs.  A few
    libs are not read because they, by design, produce output
    when read even when config("lib_debug") is set to -1.

    Increased the maximum number of object types that one can define
    from 10 to 128.

    Added a patch from Ernest Bowen <ernie at turing dot une dot edu
    dot au> to correctly hash a V_STR value-type that has an \0 byte
    inside it.

    A patch from Ernest Bowen <ernie at turing dot une dot edu dot au> now
    defines special meaning to the first 2 bits of config("lib_debug"):

	bit 0 set => messages printed when inputisterminal
	bit 1 set => messages printed when reading from a file

    The lib/regress.cal regression suite does:

	config("lib_debug", -4);

    to eliminate lib messages (both bit 0 and bit 1 are not set).

    Fixed misc compile warnings and notices.


The following are the changes from calc version 2.10.3t5.38 to 2.11.0t0:

    Fixed a few compile problems found under Red Hat 6.0 Linux.


The following are the changes from calc version 2.10.3t5.38 to 2.11.3t5.46:

    Fixed a bug discovered by Ernest Bowen related to matrix-to-matrix copies.

    Bitwise operations on integers have been extended so that negative
    integers are treated in the same way as the integer types in C.

    Some changes have been made to lib/regress.cal and lib/natnumset.cal.

    Removed V_STRLITERAL and V_STRALLOC string type constants and
    renumbered the V_protection types.

    Added popcnt(x, bitval) builtin which counts the number of
    bits in x that match bitval.

    Misc compiler warning fixes.

    Fixed improper use of putchar() and printf() when printing rationals
    (inside qio.c).

    Fixed previously reported bug in popcnt() in relation to . values.

    Calc man page changes per suggestion from Martin Buck
    <Martin-2.Buck at student dot uni-ulm dot de>.  The calc man page is
    edited with a few more parameters from the Makefile.

    Misc Makefile changes per Martin Buck <Martin-2.Buck at student dot
    uni-ulm dot de>.

    Removed trailing blanks from files.

    Consolidated in the Makefile, where the debug and check rules are found.
    Fixed the regress.cal dependency list.

    Make chk and check will exit with an error if check.awk detects
    a problem in the regression output.	 (Martin Buck)

    Fixed print line for test #4404.

    Moved custom.c and custom.h to the upper level to fix unresolved symbols.

    Moved help function processing into help.c.

    Moved nearly everything into libcalc.a to allow programs access to
    higher level calc objects (e.g., list, assoc, matrix, block, ...).

    Renamed PATCH_LEVEL to MAJOR_PATCH and SUB_PATCH_LEVEL to MINOR_PATCH.
    Added integers calc_major_ver, calc_minor_ver, calc_major_patch
    and string calc_minor_patch to libcalc.a.  Added CALC_TITLE to hold
    the "C-style arbitrary precision calculator" string.

    The function version(), now returns a malloced version string
    without the title.

    Consolidated multiple SGI IRIX -n32 sections (for r4k, r5k and r10k)
    into a single section.




The following are the changes from calc version 2.10.3t5.34 to 2.10.3t5.37:

    Per request from David I Bell, the README line:

      I am allowing this calculator to be freely distributed for personal uses

    to:

      I am allowing this calculator to be freely distributed for your enjoyment

    Added help files for:

	address agd arrow dereference free freeglobals freeredc freestatics
	gd isptr mattrace oldvalue saveval & * -> and .

    Fixed blkcpy() and copy() arg order and processing.	 Now:

	A = blk() = {1,2,3,4}
	B = blk()
	blkcpy(B,A)
	blkcpy(B,A)

    will result in B being twice as long as A.

    Since "make chk" pipes the regression output to awk, we cannot
    assume that stdout and stderr are ttys.  Tests #5985 and #5986
    have been removed for this reason.	(thanks to Martin Buck
    <Martin-2.Buck at student dot uni-ulm dot de> for this report)

    Fixed the order of prints in regress.cal.  By convention, a print
    of a test line happens after the test.  This is because function
    parsed messages occur after the function is parsed.	 Also the
    boolean test will verify before any print statements.  Therefore
    a non-test line is tested and printed as follows:

	y = sha();
	print '7125: y = sha()';

    The perm(a,b) and comb(a,b) have been extended to arbitrary real a and
    integer b.

    Fixed a bug in minv().

    Moved string.c into libcalc.a.

    The NUMBER union was converted back into a flat structure.	Changes
    where 'num' and 'next' symbols were changed to avoid #define conflicts
    were reverse since the #define's needed to support the union went away.

    Removed trailing blanks from files.

    Ernest Bowen <ernie at turing dot une dot edu dot au> sent in the
    following patch which is described in the next 34 points:

    (0) In the past:

		    A = B = strcat("abc", "def");

	would store "abc" and "def" as literal strings never to be freed, and
	store "abcdef" once each for both A and	 B.  Now the "abc" and "bcd"
	are freed immediately after they are concatenated and "abcdef" is stored
	only once, just as the number 47 would be stored only once for

		    A = B = 47;

	The new STRING structure that achieves this stores not only the
	address of the first character in the string, but also the "length"
	with which the string was created, the current "links" count, and
	when links == 0 (which indicates the string has been freed) the
	address of the next freed STRING.  Except for the null string "",
	all string values are "allocated"; the concept of literal string
	remains for names of variables, object types and elements, etc.

    (1) strings may now include '\0', as in A = "abc\0def".  In normal printing
	this prints as "abc" and strlen(A) returns 3, but its "real" length
	of 7 is given by size(A). (As before there is an 8th zero character
	and sizeof(A) returns 8.)

    (2) If A is an lvalue whose current value is a string of size n, then
	for 0 <= i < n, A[i] returns the character with index i as an addressed
	octet using the same structure as for blocks, i.e. there is no
	distinction between a string-octet and a block-octet.  The same
	operations and functions can be used for both, and as before, an octet
	is in some respects a number in [0,256) and in others a one-character
	string.	 For example, for A = "abc\0def" one will have both A[0] == "a"
	and A[0] == 97.	 Assignments to octets can be used to change
	characters in the string, e.g. A[0] = "A", A[1] = 0, A[2] -= 32,
	A[3] = " " will change the above A to "A\0C def".

    (3) "show strings" now displays the indices, links, length, and some or all
	of the early and late characters in all unfreed strings which are values
	of lvalues or occur as "constants" in function definitions,
	using "\n", "\t", "\0", "\252", etc. when appropriate.	For example,
	the string A in (1) would be displayed as in the definition there.
	Only one line is used for each string.	I've also changed the
	analogous "show numbers" so that only some digits of numbers that
	would require more than one line are displayed.

    (4) "show literals" is analogous to "show constants" for number "constants"
	in that it displays only the strings that have been introduced by
	literal strings as in A = "abc".  There is a major difference between
	strings and numbers in that there are operations by which characters
	in any string may be changed.  For example, after A = "abc",
	A[0] = "X" changes A to "Xbc".	It follows that if a literal string
	is to be constant in the sense of never changing, such a character-
	changing operation should never be applied to that string.

	In this connection, it should be noted that if B is string-valued, then

			    A = B

	results in A referring to exactly the same string as B rather than to
	a copy of what is in B.	 This is like the use of character-pointers in
	C, as in

			    char *s1, *s2;
			    s1 = "abc";
			    s2 = s1;

	To achieve the effect of

			    s2 = (char *) malloc(4);
			    strcpy(s2, s1);

	I have extended the str() function to accept a string as argument.  Then

			    A = str(B);

	will create a new string at a different location from that of B but
	with the same length and characters.  One will then have A == B,
	*A == *B, but &*A != &*B, &A[0] != &B[0].

	To assist in analyzing this sort of thing, I have defined a links()
	function which for number or string valued argument returns the number
	of links to the occurrence of that argument that is being referred to.
	For example, supposing "abc" has not been used earlier:

			    ; A = "abc"
			    ; links(A)
				    2
			    ; links(A)
				    1

	The two links in the first call are to A and the current "oldvalue";
	in the second call, the only link is to A, the oldvalue now being 2.


    (5) strcat(S1, S2, ...) works as before; contribution of a string stops when
	'\0' is encountered.  E.g.

		    strcat("abc\0def", "ghi")

	will return "abcghi".

    (6) For concatenation of full strings I have chosen to follow
	some other languages (like Java, but not Mathematica which uses "<>")
	and use "+" so that, e.g.

		    "abc\0def" + "ghi"

	returns the string "abc\0defghi".  This immediately gives obvious
	meanings to multiplication by positive integers as in

		    2 * "abc" = "abc" + "abc" = "abcabc",

	to negation to reverse as string as in

		    - "abc" = "cba",

	to multiplication by fractions as in

		    0.5 * "abcd" = "ab",

	(where the length is int(0.5 * size("abcd")), and finally, by combining
	these to

		     k * A    and      A * k

	for any real number k and any string A.	  In the case of k == 1, these
	return a new string rather than A itself.  (This differs from
	"" + A and A + "" which return A.)

    (7) char(x) has been changed so that it will accept any integer x or octet
	as argument and return a string of size one with character value
	x % 256.  In the past calc has required 0 <= x < 256; now negative
	x is acceptable; for example, 1000 * char(-1) will now produce the
	same as 1000 * "\377" or 1000 * "\xff".

    (8) For a string s, test(s) now returns zero not only for the null string
	"" but also for a string all of whose characters are '\0'.

    (9) Similarly <, <=, etc. now compare all characters including occurrences
	of '\0' until a difference is encountered or the end of a string is
	reached.  If no difference is encountered but one string is longer than
	the other, the longer string is considered as greater even if the
	remaining characters are all '\0'.

    (10) To retain the C sense of comparison of null-terminated strings I have
	 defined strcmp(S1, S2), and then, for completeness, strncmp(S1, S2, n).
	 For similar reasons, strcpy(S1, S2) and strncpy(S1, S2, n) have been
	 defined.

    (11) For strings, I have defined | and & as bitwise "or" and "and"
	 functions, with S1 | S2 having the size of the larger of S1 and S2,
	 S1 & S2 having the size of the smaller of S1 and S2.  By using, say,
	 4-character strings, one can simulate a C integral type so far as the
	 | and & operations are concerned.   It then seemed appropriate to
	 use the operator ~ for a "bitwise complement" as in C.	 Thus I have
	 defined ~s for a string s to be the string of the same size as s
	 with each character being complemented by the C ~ operation.

    (12) For boolean algebra work on strings it is convenient also to have
	 the bitwise xor and setminus binary operations.  Using C's '^' for xor
	 would be confusing when this is used elsewhere for powers, so I
	 decided to use ~.  For setminus, I adopted the commonly used '\'.
	 Strings of fixed size n can now be used for a boolean algebra
	 structure with 8 * n elements.	 The zero element is n * char(0),
	 the unity is n * char(-1), and one have all of the usual laws like
	 A & (B | C) == A & B | A * C,	A \ B = A & ~B, etc.

    (13) Having extended the bitwise operations for strings, it was appropriate
	 to do the same for integers.  Definitions of the binary ~ and \
	 operations for non-negative integers are straightforward.  For
	 the unary ~ operation, I decided to do what C does with integer
	 types, and defined ~N to be -N - 1.  With the appropriate extensions of
	 |, &, \ and the binary ~, one gets in effect the boolean algebra of
	 finite sets of natural numbers and their complements, by identifying
	 the set with distinct integer elements i_1, i_2, ... with the integer

		    2^i_1 + 2^i_2 + ...

	 For ~N for non-integer real N, I have simply used -N.	There is some
	 logic in this and it is certainly better than an error value.
	 I have not defined the binary operations |, &, ~, \ for non-integral
	 arguments.

	 The use of ~N in this way conflicts with calc's method of displaying
	 a number when it has to be rounded to config("display") decimals.
	 To resolve this, my preference would be to replace the printing of
	 "~" as a prefix by a trailing ellipsis "...", the rounding always
	 being towards zero.  E.g. with config("display", 5), 1/7 would print
	 as ".14285..." rather than "~.14285".	 The config("outround")
	 parameter would determine the type of rounding only for the
	 equivalent of config("tilde", 0).

    (14) For objects, users may create their own definitions for binary |,
	 &, ~ and \ with xx_or, xx_and, xx_xor, xx_setminus functions.
	 For unary ~ and \ operations, I have used the names xx_comp and
	 xx_backslash.

    (15) For the obviously useful feature corresponding to cardinality of a
	 set, I have defined #S for a string S to be the number of nonzero bits
	 in S.	 For a degree of consistency, it was then appropriate to
	 define #N for a nonnegative integer N to be the number of nonzero bits
	 in the binary representation of N.  I've extended this to arbitrary
	 real N by using in effect #(abs(num(N))).  I feel it is better to make
	 this available to users rather than having #N invoke an error message
	 or return an error value.  For defining #X for an xx-object X, I
	 have used the name xx_content to suggest that it is appropriate for
	 something which has the sense of a content (like number of members of,
	 area, etc.).

    (16) Having recognized # as a token, it seemed appropriate to permit its
	 use for a binary operation.  For real numbers x and y I have defined
	 x # y to be abs(x - y).  (This is often symbolized by x ~ y, but it
	 would be confusing to have x ~ y meaning xor(x,y) for strings and
	 abs(x-y) for numbers.)	 Because '#' is commonly called the hash symbol,
	 I have used xx_hashop to permit definition of x # y for xx-objects.

    (17) For a similar reason I've added one line of code to codegen.c so that
	 /A returns the inverse of A.

    (18) Also for a list L, +L now returns the sum of the elements of L.  For
	 an xx object A, +A requires and uses the definition of xx_plus.

    (19) I have given the unary operators ~, #, /, \, and except at the
	 beginning of an expression + and -, the same precedence with
	 right-to-left associativity.  This precedence is now weaker than
	 unary * and &, but stronger than binary & and the shift and power
	 operators.  One difference from before is that now

			    a ^ - b ^ c

	 evaluates as a ^ (- (b ^ c)) rather than a ^ ((- b) ^ c).


    (20) For octets o1, o2, I've defined o1 | o2, o1 & o2, o1 ~ o2, ~o1 so
	 that they return 1-character strings.	#o for an octet o returns the
	 number of nonzero bits in o.

    (21) For substrings I've left substr() essentially as before, but
	 for consistency with the normal block/matrix indexing, I've extended
	 the segment function to accept a string as first argument.  Then

		    segment(A, m, n)

	 returns essentially the string formed from the character with index m
	 to the character with index n, ignoring indices < 0 and indices >=
	 len(A); thus, if m and n are both in [0, size(A))
	 the string is of length abs(m - n) + 1, the order of the characters
	 being reversed if n < m.  Here the indices for a list of size len are
	 0, 1, ..., len - 1.  As it makes some sense, if 0 <= n < size(A),

		    segment(A, n)

	 now returns the one-character string with its character being that with
	 index n in A.	(I've made a corresponding modification to the segment
	 function for lists.)  Some examples, if A = "abcdef",

		    segment(A,2,4) = "cde",

		    segment(A,4,2) = "edc",

		    segment(A,3) = "d",

		    segment(A, -2, 8) = "abcdef",

		    segment(A,7,8) = "".

    (22) As essentially particular cases of segment(), I've defined
	 head(A, n) and tail(A, n) to be the strings formed by the first
	 or last abs(n) characters of A, the strings being4]5O~? reversed '
	 if n is negative.   I've changed the definitions of head and tail for
	 lists to be consistent with this interpretation of negative n.

    (23) Similarly I've left strpos essentially as at present, but search
	 and rsearch have been extended to strings.  For example,

		    search(A, B, m, n)

	 returns the index i of the first occurrence of the string B in A
	 if m <= i < n, or the null value if there is no such occurrence.
	 As for other uses of search, negative m is interpreted as
	 size(A) + m, negative n as size(A) + n.  For a match in this
	 search, all size(B) characters, including occurrences of '\0',
	 in B must match successive characters in A.

	 The function rsearch() behaves similarly but searches in reverse order
	 of the indices.

    (24) A string A of length N determines in obvious ways arrays of M = 8 * N
	 bits.	If the characters in increasing index order are c_0, c_1, ...
	 and the bits in increasing order in c_i are b_j, b_j+1, ..., b_j+7
	 where j = 8 * i, I've taken the array of bits determined by A to be

		    b_0, b_1, ..., b_M-1

	 For example, since "a" = char(97) and 97 = 0b01100001, and
	 "b" = char(98) = 0b01100010, the string "ab" determines the 16-bit
	 array

		    1000011001000110

	 in which the bits in the binary representations of "a" and "b" have
	 been reversed.

	 bit with index n in this array.   This is consistent with the use of
	 bit for a number ch in [0,256), i.e. bit(char(ch), n) = bit(ch, n).
	 For n < 0 or n >= size(A), bit(A,n) returns the null value.

    (25) For assigning values to specified bits in a string, I've defined
	 setbit(A, n) and setbit(A, n, v).  The first assigns the value 1 to
	 bit(A, n), the second assigns test(v) to bit(A, n).

    (26) For consistency with the corresponding number operations, the shift
	 operations A << n and A >> n have been defined to give what look
	 like right- and left-shifts, respectively.  For example, "ab" << 2
	 returns the 16-bit array

		    0010000110010001

	 in which the array for "ab" has been moved 2 bits to the right.

    (27) To achieve much the same as the C strcpy and strncpy functions for
	 null-terminated strings, strcpy(S1, S2) and strncpy(S1, S2, n) have
	 been defined.	Unlike the blkcpy() and copy() functions, the copying
	 for these is only from the beginning of the strings.  Also, unlike C,
	 no memory overflow can occur as the copying ceases when size(S1) is
	 reached.  Note that these overwrite the content of S1 (which affects
	 all strings linked to it) as well as returning S1.  Examples:

	    S = strcpy(6 * "x", "abc")	    <=>	 S = "abc\0xx"

	    S = strcpy(3 * "x", "abcdef")   <=>	 S = "abc"

	    S = strncpy(6 * "x", "abcd", 2) <=>	 S = "ab\0xxx"

	    S = strncpy(6 * "x", "ab", 4)   <=>	 S = "ab\0\0xx"

	    S = strncpy(6 * "x", "ab", 20)  <=>	 S = "ab\0\0\0\0"

	 If a new string S not linked to S1 is to be created, this can be
	 achieved by using str(S1) in place of S1.  For example, the strcpy in

	    A = "xxxxxx"
	    S = strcpy(str("xxxxxx"), "abc")

	 would not change the value of A.

    (28) I've extended the definitions of copy(A, B, ssi, num, dsi) and
	 blkcpy(B, A, num, ssi, dsi) to allow for string-to-string copying
	 and block-to-string copying, but num is now an upper bound for the
	 number of characters to be copied - copying will cease before num
	 characters are copied if the end of the data in the source A or the
	 end of the destination B is reached.  As with other character-changing
	 operations, copying to a string B will not change the locations of
	 B[0], B[1], ... or the size of B.

	 In the case of copying a string to itself, characters are copied in
	 order of increasing index, which is different from block-to-block
	 copying where a memmove is used.  This affects only copy from a
	 string to itself.  For example,

		    A = "abcdefg";
		    copy(A, A, , , 2);

	 will result in A == "abababa".	 If the overwriting that occurs here
	 is not wanted, one may use

		    A = "abcdefg";
		    copy(str(A), A, , , 2);

	  which results in A == "ababcde".

    (29) perm(a,b) and comb(a,b) have been extended to accept any real a and
	 any integer b except for perm(a, b) with integer a such that b <= a < 0
	 which gives a "division by zero" error.  For positive b, both functions
	 are polynomials in a of degree b;  for negative b, perm(a,b) is a
	 rational function (1/((a + 1) * (a+2) ...) with abs(b) factors in the
	 denominator), and comb(a,b) = 0.  (An obvious "todo" is to extend this
	 to complex or other types of a.)

    (30) Although it is not illegal, it seems pointless to use a comma operator
	 with a constant or simple variable as in

		    ; 2 * 3,14159
			    14159
		    ; a = 4; b = 5;
		    ; A = (a , b + 2);
		    ; A
			    7

	 I have added a few lines to addop.c so that when this occurs a
	 "unused value ignored" message and the relevant line number are
	 displayed.  I have found this useful as I occasionally type ','
	 when I mean '.'.

	 There may be one or two other changes resulting from the way I have
	 rewritten the optimization code in addop.c.  I think there was a bug
	 that assumed that PTR_SIZE would be the same as sizeof(long).	By
	 the way, the new OP_STRING is now of index rather than pointer type.
	 It follows that pointers are now used in opcodes only for global
	 variables.  By introducing a table of addresses of global variables
	 like those used for "constants" and "literal strings", the use of
	 pointers in opcodes could be eliminated.

    (31) When calc has executed a quit (or exit) statement in a function or
	 eval evaluation, it has invoked a call to math_error() which causes
	 a long jump to an initial state without freeing any data on the
	 stack, etc.  Maybe more detail should be added to math_error(), but
	 to achieve the freeing of memory for a quit statement and at the same
	 time give more information about its occurrence I have changed the
	 way opcodes.c handles OP_QUIT.	 Now it should free the local variables
	 and whatever is on the stack, and display the name and line-number,
	 for each of the functions currently being evaluated.  The last
	 function listed should be the "top-level" one with name "*".
	 Strings being eval-ed will have name "**".

	 Here is a demo:

	    ; global a;
	    ;
	    ; define f(x) {local i = x^2; a++;
	    ;; if (x > 5) quit "Too large!"; return i;}
	    f() defined
	    ; define g(x) = f(x) + f(2*x);
	    g() defined
	    ; g(2)
		    20
	    ; g(3)
	    Too large!
		    "f": line 3
		    "g": line 0
		    "*": line 6
	    ; eval("g(3)")
	    Too large!
		    "f": line 3
		    "g": line 0
		    "**": line 1
		    "*": line 7
	    ; a
		    6

    (32) I've made several small changes like removing

		    if (vp->v_type == V_NUM) {
			    q = qinv(vp->v_num);
			    if (stack->v_type == V_NUM)
				    qfree(stack->v_num);
			    stack->v_num = q;
			    stack->v_type = V_NUM;
			    return;
		    }

	 from the definition of o_invert.  Presumably these lines were intended
	 to speed up execution for the common case of numerical argument.
	 Comparing the runtimes with and without these lines for inverting
	 thousands of large random numbers in a matrix suggest that execution
	 for real numbers is slightly faster without these lines.

	 Maybe this and other similar treatment of "special cases" should be
	 looked at more closely.

    (33) The new lib script lib/natnumset.cal demonstrates how the new
	 string operators and functions may be used for defining and
	 working with sets of natural numbers not exceeding a
	 user-specified bound.


The following are the changes from calc version 2.10.3t5.28 to 2.10.3t5.33:

    Added hnrmod(v, h, n, r) builtin to compute:

	v % (h * 2^n + r), h>0, n>0, r = -1, 0 or 1

    Changed lucas.cal and mersenne.cal to make use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(z), i.e. the gudermannian and
	    inverse gudermannian

	(6) introduction of show options for displaying information about
	    current constants, global variables, static variables, and cached
	    redc moduli.

	    To help you follow what is going on, the following show
	    items have been introduced:

		show constants ==> display the currently stored constants
		show numbers   ==> display the currently stored numbers
		show redcdata  ==> display the currently stored redc moduli
		show statics   ==> display info about static variables
		show real      ==> display only real-valued variables

	    The constants are automatically initialized as constants and
	    should always appear, with links >= 1, in in the list of constants.

	    The show command:

		show globals

	    has been redefined so that it gives information about all
	    current global and still active static variables.

	(7) definition of functions for freeing globals, statics, redc values

	    To free memory used by different kinds of variable, the following
	    builtins have been added:

		freeglobals();		/* free all globals */
		freestatics();		/* free all statics */
		freeredc();		/* free redc moduli */
		free(a, b, ...);	/* free specific variables */

	   NOTE: These functions do not "undefine" the variables, but
	   have the effect of assigning the null value to them, and so
	   frees the memory used for elements of a list, matrix or object.

	   See 10) below for info about "undefine *".

	(8) enhancement of handling of "old value": having it return an
	    lvalue and giving option of disabling updating.

	    Now, by default, "." return an lvalue with the appropriate
	    value instead of copying the old value.

	    So that a string of commands may be given without changing
	    the "oldvalue", the new builtin:

		saveval(0)

	    function simply disables the updating of the "." value.
	    The default updating can be resumed by calling:

		saveval(1)

	    The "." value:

		; 2 + 2
		4
		; .
		4

	    can now be treated as an unnamed variable.	For example:

		; mat x[3,3]={1,2,3,4,5,6,7,8,9}
		; x
		; print .[1,2]
		6

	(9) for a list L defining L[i] to be same as L[[i]]

	(10) extending undefine to permit its application to all user-defined
	     functions by using "undefine *".

	     The command:

		undefine *

	     undefines all current user-defined functions.  After
	     executing all the above freeing functions (and if
	     necessary free(.) to free the current "old value"), the
	     only remaining numbers as displayed by:

		show numbers

	     should be those associated with epsilon(), and if it has been
	     called, qpi().

	(11) storing the most recently calculated value of qpi(epsilon)i and
	     epsilon so that when called again with the same epsilon it
	     is copied rather than recalculated.

	(12) defining trace() for square matrices

	(13) expression in parentheses may now be followed by a qualifier
	     computable with its type

	     When an expression in parentheses evaluates to an lvalue
	     whose current value is a matrix, list or object, it may
	     now be followed by a qualifier compatible with its type.

	     For example:

		; A = list(1,2,4);
		; B = mat[2,2] = {5,6,7,8};
		; define f(x) = (x ? A : B)[[1]];
		; print f(1), f(0)
		2 6

		; obj xy {x,y}
		; C = obj xy = {4,5}
		; p = &C
		; *p.x
		Not indexing matrix or object
		; (*p).x
		4

	(14) swap(a,b) now permits swapc use of hnrmod().

    A number of changes from Ernest Bowen:

	(1) introduction of unary & and * analogous to those in C;

	    For an lvalue var, &var returns what I call a
	    value-pointer; this is a constant which may be assigned to
	    a variable as in p = &var, and then *p in expressions has
	    the same effect as var.  Here is a simple example of their use:

		; define s(L) {local v=0; while (size(L)) v+= *pop(L);return v;}
		s() defined
		; global a = 1, b = 2;
		; L = list(&a, &b);
		; print s(L)
		3
		; b = 3;
		; print s(L)
		4

	    Octet-pointers, number-pointers, and string-pointers in
	    much the same way, but have not attempted to do much with
	    the latter two.

	    To print a pointer, use the "%p" specifier.

	    Some arithmetic operations has been defined for corresponding
	    C operations.  For example:

		; A = mat[4];
		; p = &A[0];
		; *(p+2) == A[2]
		; ++p
		; *p == A[1]

	    There is at present no protection against "illegal" use of &
	    and *, e.g. if one attempts here to assign a value to *(p+5),
	    or to use p after assigning another value to A.

	    NOTE: Unlike C, in calc &A[0] and A are quite different things.

	    NOTE: If the current value of a variable X is an octet,
	    number or string, *X may be used to to return the value of
	    X; in effect X is an address and *X is the value at X.

	    Added isptr(p) builtin to return 0 is p is not a pointer,
	    and >0 if it is a pointer.  The value of isptr(p) comes from
	    the V_XYZ #define (see the top of value.h) of the value to
	    which p points.

	    To allow & to be used as a C-like address operator, use of it
	    has been dropped in calls to user-defined functions.  For the
	    time being I have replaced it by the back-quote `.	For example:

		; global a
		; define f(a,b) = a = b
		; f(&a,5)
		; print a
		0
		; f(`a,5)
		; print a
		5

	   However, one may use & in a similar way as in:

		; define g(a,b) = *a = b
		; g(&a, 7)
		; print a
		7

	   There is no hashvalue for pointers. Thus, like error values,
	   they cannot be used as indices in an association.

	   The -> also works in calc. For example:

		; obj xy {x,y}
		; obj uvw {u, v, w}
		; obj xy A = {1,2}
		; obj uvw B = {3,4,5}
		; p = &A
		; q = &B
		; p->x
			1
		; p->y = 6
		; A
			obj xy {1, 6}
		; q -> u
			3
		; p->y = q
		; A
			obj xy {1, v-ptr: 1400474c0}
		; p->y->u
			3
		; p->y->u = 7
		; B
			obj uvw {7, 4, 5}
		; p -> y = p
		; A
			obj xy {1, v-ptr: 140047490}
		; p -> y -> x
			1
		; p->y->y
			v-ptr: 140047490
		; p->y->y-> x
			1
		; p->y->y->x = 8
		; A
			obj xy {8, v-ptr: 140047490}


	(2) a method of "protecting" variables;

	    For the various kinds of "protection", of an l_value var,
	    bits of var->v_subtype, of which only bits 0 and 1 have been
	    used in the past to indicate literal and allocated strings.
	    This has meant initialization of var->v_subtype when a new var
	    is introduced, and for assignments, etc., examination of the
	    appropriate bits to confirm that the operation is to be permitted.

	    See help/protect for details.

	(3) automatic "freeing" of constants that are no longer required.

	    For the "freeing" of constants, the definition of a NUMBER
	    structure so that a NUMBER * q could be regarded as a
	    pointing to a "freed number" if q->links = 0.

	    The old q->num was changed to a union q->nu which had a pointer
	    to the old q->num if q->links > 0 and to the next freed number
	    if q->links = 0.  The old "num" is #defined to "nu->n_num".

	    The prior method calc has used for handling "constants" amounted
	    to leakage.	 After:

		; define f(x) = 27 + x;
		; a = 27;

	    It is of course necessary for the constant 27 to be stored, but
	    if one now redefines f and a by:

		; define f(x) = 45 + x;
		; a = 45;

	    There seems little point in retaining 27 as a constant and
	    therefore using up memory.	If this example seems trivial,
	    replace 27 with a few larger numbers like 2e12345, or better,
	    -2e12345, for which calc needs memory for both 2e12345 and
	    -2e12345!

	    Constants are automatically freed a definition when a
	    function is re- or un-defined.

	    The qalloc(q) and qfree(q) functions have been changed so
	    that that q->links = 0 is permitted and indicates that q
	    has been freed.  If a number has been introduced as a
	    constant, i.e. by a literal numeral as in the above
	    examples, its links becoming zero indicates that it is no
	    longer required and its position in the table of constants
	    becomes available for a later new constant.

	(4) extension of transcendental functions like tan, tanh, etc.
	    to complex arguments

	(5) definition of gd(z) and agd(