LWN: Comments on "Using Guile for Emacs"

Reducing the amount of C in Emacs

ballombe — Wed, 18 Dec 2024 21:08:57 +0000

You can compile C to wasm, so it might not be that bad.

Reducing the amount of C in Emacs

nim-nim — Wed, 18 Dec 2024 16:06:38 +0000

While I wish them the best for their project if C is now “around 1/4 of the codebase” hoping to reduce its footprint to the level needed to “compile [Emacs] to WebAssembly ” seems more like a complete rewrite and probably pure wishful thinking.

Guile's platform support

epa — Wed, 18 Dec 2024 14:10:20 +0000

I think those are platforms which GNU Emacs supports (or did run on them, the last time anyone checked) but which Guile has never supported.

I'd guess the odds are pretty good that if you took a current GNU Emacs tarball you could still build and run it on your ConvexOS system, if you had one (or Alpha, or whatever). Emacs hasn't explicitly removed support for them as there was no particular need to do so. The Guile-Emacs docs are being explicit that these won't be a target for the new project.

Guile's platform support

taladar — Wed, 18 Dec 2024 09:27:59 +0000

Agreed. Platform support shouldn't even be considered for platforms where you don't have at least one developer with access to the platform to perform tests on that platform. I would have thought that should be obvious but it seems some of these projects bend over backwards for obscure niche platforms that last fulfilled that requirement decades ago (or never did as in that Nonstop platform).

Guile's platform support

willy — Tue, 17 Dec 2024 22:46:28 +0000

How would anyone know if ConvexOS support is broken? AFAICT no hardware has shipped running that OS since 1995. I doubt you can find a system simulator for it. The machines that ran it were measured in cubic metres (and kilowatts of power).

I'm all for fun retrocomputing, but supporting necrocomputing is not something that should hold any project back (see recent article on Rust, Git and Nonstop)

Guile's platform support

tome — Tue, 17 Dec 2024 15:03:40 +0000

See https://codeberg.org/guile-emacs/guile-emacs#headline-31

Guile's platform support

tome — Tue, 17 Dec 2024 15:01:33 +0000

The README file in guile-emac's Codeberg repo has a partial list of platforms unsupported by guile, including

> Ardent, Ultrix, ConvexOS, OpenIndiana, 32-bit OS on SPARC64, GNU/Linux Alpha

and mentions a two-layer effort for guilemacs:

> * guile-emacs that continues to rebase ontop of vanilla emacs.

> * opinionated guilemacs, that builds furhter ontop of guile-emacs, removes unwanted stuff, less target platforms, enabler for more radical ideas.

so, basically soft and hard forks of emacs, with the hard fork forsaking many platforms.

Xemacs was a hard fork of Emacs created to do things Emacs didn't and wouldn't do, which lasted until Emacs eventually did do those things too. If "opinionated guilemacs" takes off with enough enthusiasm, one can imagine a similar process playing out over a period of years.

Guile's platform support

gray_-_wolf — Tue, 17 Dec 2024 11:45:13 +0000

I have to admit that as much as a like this project, and would love it to succeed, I am bit concerned about multi-platform support. I *think* Guile currently does not run on Windows for example (well, it does in cygwin, sure, but that can hardly be the goal), so I wonder what are the plans here.

Some Emacs extensions contain C code

fw — Mon, 16 Dec 2024 21:08:06 +0000

In case anyone is confused by C extensions for Emacs, I think it's about this: https://www.gnu.org/software/emacs/manual/html_node/elisp...

guile-emacs was impressive and still is

paroneayea — Mon, 16 Dec 2024 18:29:51 +0000

I am glad to hear of Robin's guile-emacs revival effort and that it's getting relevant attention. guile-emacs is an impressive and frankly incredible project.

I think people may be surprised just how much works of it too. In 2015, I took a screenshot of me running it with my config file working, my theme, and an org-mode buffer and some elisp both open: https://dustycloud.org/tmp/guile-emacs-20150513.png

I've long felt sad that guile-emacs was sitting on the shelf. I was sad that much of the emacs community gave guile-emacs such a lukewarm reception a decade ago. I hope that can change today.

At the time, guile-emacs was mostly unoptimized, but the optimization work needed was known. It was amazing being able to run both elisp and scheme code both between it, and with Guile's compiler tower, it would be possible to run even more languages. Guile has also received a huge amount of work on performance which emacs could itself benefit from; another interesting integration point could be the work on Scheme-to-WASM with Hoot (which Robin has contributed significantly to also). I'm not sure how that latter part would play out, but it's no doubt interesting.

So all I will say to wrap this up is: go guile-emacs! I hope the project succeeds.

Some Emacs extensions contain C code

paroneayea — Mon, 16 Dec 2024 18:20:04 +0000

Note that the lead author of porting Emacs Lisp to Guile was Robin Templeton themselves, mentored by Andy Wingo.

Some Emacs extensions contain C code

tome — Mon, 16 Dec 2024 15:56:41 +0000

Back when Andy Wingo finished developing a complete implementation of elisp in Guile (about 15 years ago!), the biggest impediment to Guile Emacs was due to the many Emacs extensions containing C code dependent on C code in Emacs that would be removed from Guile Emacs. It would take a mountain of work to migrate those extensions to Guile Emacs. That's my recollection. But this article mentions that

> A year after their last GSoC project, they had a Guile-Emacs prototype that "was completely compatible with Emacs functionality and with external extensions".

I don't know enough about Emacs internals or extensions in C to sort out this contradiction. Without the C extensions roadblock it seems the advantages of Guile Emacs would be overwhelming enough to overcome any merely political objections.