Updated 2009-02-09 17:31:25 by LV ▲

Bignum v1.0 - a critcl tcl wrapper to libgmp
WHAT
provides these packages:

  • bigint - which provides wrappers for arbitrary precision integers
  • bigfloat - which provides wrappers for arbitrary precision floats
  • bignum - both bigint and bigfloat
WHERE
http://sourceforge.net/project/showfiles.php?group_id=18733
DEPENDENCIES
This works with libgmp3 4.1.1 or similar under Linux; get it from any gnu archive (if you haven't got it). To recompile it, you'll need the latest version of critcl; http://tcl.tk/starkits/ is where you'll find it.
INSTALL
unpack this in /usr/lib, /usr/local/lib, or wherever your tcl resides.
AUTHOR
Colin McCormack
LICENSE
It's licensed under GPL - of necessity, as libgmp is.

SEE ALSO: [1] SS's bignum in pure Tcl

I've modified bignum to use the latest Critcl and built a cross-platform library Starkit with binaries for Linux, Windows and OSX [2]

To use, either unpack the Starkit and copy the contents of the lib directory into somewere on your auto_path, or just source it directly
   tclkit
   % source bignum.kit
   % package require bignum
   % bigint::add 1111111111111111111111111111 2222222222222222222222222222
   3333333333333333333333333333

The binaries on Windows and Linux are statically linked against libgmp (and therefore self-contained) but I couldn't get this working on OSX so you'll need to have gmp installed on your LDPATH to use bignum on OSX. I'll have another go later.

The modified source is available at [3]

stevel 27 Jan 05

LES on July 27 2005: using bignum.kit from sdarchive with ActiveTcl 8.5 on Windows 98, I ran this command...
 $ ::bigint::powm 5 6 0

...and Tcl crashed (pure DOS or Tkcon). It only seems to happen with the 0, but it does happen every time. Maybe I shouldn't be using 0, I admit I do not understand what that function really does (I wanted to calculate a large number raised to a large power). But I don't think that Tcl or bignum should crash on me just because I threw a zero at it. :-)

Lars H: powm is probably "power modulo" (a very common operation in many cryptographical algorithms), so a zero as last argument is likely to lead to "divide by zero" conditions. This could either be a bug in libgmp, or a failure of the wrapper to catch input not supported by that library, but come Tcl 8.5 we'll most likely have equivalent functionality in the core anyway (happy, happy; joy, joy).

This [4] is a review of a book called BigNum Math.

Category Package Category Mathematics