Re: [Maxima-discuss] Trying to accommodate GCL is keeping us from making progress

[Robert S. <ro...@st...>]

Dear RD:

I am in full support for eliminating support for GCL. I am happy to commit time and energy to excising it in a clean, reviewable pull request/patch on GitHub.

With that said, here are my good-faith beliefs about why supporting GCL is a good thing. (I do *not* think the totality of reasons is worth keeping it, however.)

Pros for keeping GCL:

- Supporting GCL means increased portability of Maxima

- - Some operating systems have GCL accessible due to it being a GNU project.

- - GCL is an older Lisp and therefore works on older systems.

- - GCL, like CLISP, gives you an inalienable right to its source code.

- Since Maxima was co-developed with GCL, Maxima may be faster in some smaller arenas (symbol-plists IIRC?) than other Lisps. (This knowledge may be out of date.)

Cons of keeping GCL:

- I'm a professional Common Lisp programmer of more than a decade, and Maxima code is very hard to decipher. Supporting GCL and its quirks only makes it more difficult. Every time I see #+gcl (compile load execute) #-gcl (:compile-toplevel ...), I wince.

- GCL's last release was 7 years ago. This is not OK for a Lisp implementation, because operating systems change, and de facto Lisp standards change.

- GCL's last commits were December 2021. The commits were small in nature.

- If I'm reading the git logs correctly, GCL is supported by exactly one person, and this person has been studiously keeping it afloat, alone, for 10+ years. GCL is not contributed to by the larger community, and, to my knowledge, has no commercial backing.

- GCL, even after a push for it, is still not fully ANSI compliant. It's the one thing it ought to have got by now.

- The larger Common Lisp community—at least the subset that makes open source libraries and distributes them—does not use GCL.

- GCL requires a C compiler.

GCL at this point is a liability to the Maxima team, Maxima contributors, and ultimately the Maxima project.

Cheers,

RS

On Mon, Oct 17, 2022 at 09:46, Robert Dodier <rob...@gm...> wrote:

> GCL has several deficiencies, and continuing to accommodate them is
> keeping us from implementing various things, aside from being a time
> sink.
>
> Things I can remember at the moment, there may be others. Also, this
> is aside from dealing with GCL-specific bugs.
>
> * wildcards in DIRECTORY -- GCL doesn't behave as expected
> * compiling pregexp.lisp -- most recent released version can't
> compile pregexp.lisp; most recent unreleased version doesn't build on
> my system
> * ASDF, UIOP, Quicklisp -- doesn't work with GCL
> * CFFI -- doesn't work with GCL
> * Unicode characters -- GCL doesn't know about that
>
> Someone's going to say, well, nobody needs any of that stuff. I guess
> that's true, but that's not very convincing, right? "We don't need
> that" is an all-purpose argument which brings everything to a dead
> halt if one really believes it.
>
> I'd like to get to a place where we can talk about the merits of this
> feature or that feature, without the conversation immediately getting
> derailed by "well, GCL doesn't support that."
>
> I know we've had similar discussions over the years. This week we have
> someone volunteering to work on Maxima, so that's what's new.
>
> At this point I'd like to discontinue support for GCL in Maxima, and
> discourage package maintainers from trying to use it, since that will
> inevitably lead to users reporting non-working features as bugs. Can I
> get anybody on board with that?
>
> FWIW
>
> Robert
>
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss