Bill Page wrote:
>
> I understand of course that your proposal is motivated by your PhD
> research.
> It seems to me however that the other fork of Axiom - the
> OpenAxiom project has for sometime had as one of its primary goals
> improvements to the SPAD compiler and the SPAD language itself. Gaby
> has even mentioned a goal of replacing Lisp as the underlying runtime
> system that sounds very similar to your work. The FriCAS project on
> the other hand has (mostly) focused on improvements to the SPAD
> library and the addition of new mathematical algorithms. Now FriCAS is
> also the only fork of Axiom that already supports Aldor as an
> alternative compiler for the FriCAS library. So my first question is:
> How does your proposal differ from what has been proposed and done so
> far in OpenAxiom and what is already available in Aldor?
I think I should comment about OpenAxiom and FriCAS here. FriCAS
improved both compiler and algebra. FriCAS fixed several bugs,
implemented new features and removed/simplified a lot of compiler
code. When I the past I compared line counts in the 'src/interp'
subdirectory for various flavours FriCAS had smallest line count.
I dare to say that was not due to code added in other flavours,
but due to most agressive cleanup in FriCAS. What FriCAS did
not do are massive renamings which were done in OpenAxiom.
Concerning Krystian's work I know that in FriCAS several
traps/unneded dpendencies from old compier are removed. Due
to renamings in OpenAxiom it is hard to give exact comparison,
but a handful of cases suggest that several problems are
still present in OpenAxiom. Of course OpenAxiom also added
features. But they does not change the fundamental structure
(problem) of old compiler. Krystian is working on quite
a different compiler structure. One thing from OpenAxiom
that could help are interfaces that expose compiler data
structures. I mentioned them to Krystian in the past,
but IIRC he found them not so helpful. I hope he can
explain this better. Maybe Gaby could give some compeling
reasons to base typechecker work on OpenAxiom, but
I recommended FriCAS for this work knowing about FriCAS
advantages and lacking knowledge about any substantial
advantages of OpenAxiom.
Concerning Aldor: typechecker structure probably will end
up beeing similar to Aldor typechecker. However, Krystian
started his work before free Aldor was available, and
it would be painful to switch direction. And Aldor
compiler is written in C, while new work is done in Spad,
so hopefully new typechecker will be shorter and simpler.
And ultimately goal is to go beyond Aldor.
Note that new typechecker will be potentially usable by
Aldor and OpenAxiom. Of course, actual use will require
substantial work. What happens will depend on merits
of new typechecker.
--
Waldek Hebisch
[email protected]
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/fricas-devel.
For more options, visit https://groups.google.com/d/optout.
