Guile-Emacs

[ludo@gnu.org (Ludovic Courtès)]

Eli Zaretskii <eliz@gnu.org> skribis:

> First, Guile's Scheme is not Emacs Lisp, there are significant
> differences.

Guile’s compiler has an Emacs Lisp front-end.

> Second, Guile itself is written in C, so what exactly is gained here?

More than half is in Scheme (see <https://www.openhub.net/p/guile>.)

The C part is a subset of the run-time support library, libguile.  The
compiler itself, among other things, is written in Scheme.

> Third, AFAIR the Guile branch doesn't replace all of Emacs's C core.

Right.  Part of Emacs’ C core becomes redundant with libguile, though,
so it should be possible to remove that part.

> Fourth, that branch is far from ready for prime time (as you know and
> point out).

This is true.  That said, of all the proposals discussed here, it seems
to be by far the one that is the closest to being ready.  :-)

<http://emacswiki.org/emacs/GuileEmacs> provides a good summary.

>> GUILE's byte compiler is supposed to do a better job than Elisp.
>
> But for now it has known problems with ELisp (some tests fail).  Also,
> at least Guile's own byte code (the *.go files) are not
> architecture-independent, so building a Guile Emacs will need a long
> compilation on the target machine.  Not a catastrophe, but hardly a
> nice thing.

There’s a trade-off here: those files can be mmap’d directly, which
reduces resource usage.

It would be wonderful if Emacs hackers could look at Guile-Emacs, talk
to Robin Templeton, identify problems, and help out.  After all the
thought and sweat Robin has put into this project, I’m confident that
it’s not (no longer!) an unrealistic project.

More importantly, the benefits are way more tangible than those of a C++
rewrite of the core; see <http://emacswiki.org/emacs/GuileEmacs>.

Let’s just do it!  :-)

Ludo’.