Re: [open-axiom-devel] other lisps
A system for computer algebra and symbolic mathematics
Brought to you by:
dos-reis
From: Waldek H. <he...@ma...> - 2007-11-28 00:13:22
|
Martin Rubey wrote: > Gabriel Dos Reis <gd...@cs...> writes: > > > On Tue, 27 Nov 2007, Gregory Vanuxem wrote: > > > > | Just out of curiosity, do you plan to support other LISP implementations > > | than GCL in the near future ? > > > > Yes, as a matter of fact, that was part of the build-improvements > > aims and the infrastructure is already there -- I `just' need more work. > > The `everything' compiled was partly motivated by the desire to use ECL. > > Since friCAS already supports SBCL, CLISP, GCL and Openmcl, wouldn't it make > sense to synchronise the relevant parts? > > Or are open-axiom and fricas already too far apart to profit from one another? "synchronise" is a bit tricky because support for SBCL, in particluar content of the src/lisp subdirectory was one of first places where build-improvements and wh-sandbox diverged in a non-trival way (non-trival meaning that the same code was changed in two different ways). OTOH OpenAxiom can easily benefit from work done in wh-sandbox and FriCAS. One easy piece is FFI support -- FriCAS FFI support is not nice (I plan to re-stucture it) but it works and can be easily re-used. Similarly platform-specific routines are easy to re-use. More generally I would say that differences outside src/interp are easily managable. In src/interp things get more tricky: OpenAxiom has 131367 lines there, FriCAs 93913. The unified diff of common files has 35272 lines. Most of differences is trivial, but bulk of the diff makes hard to spot the essential ones. I belive that FriCAS and OpenAxiom will be able for long time to "port" improvements from one to another, but "synchronizing" both codebases is more difficult. -- Waldek Hebisch he...@ma... |