Emacs and Guile status (was Re: Guile is a great idea, but where's the community?)

[Ken Raeburn]

Neil Jerram <address@hidden> writes:

>>>>>> "andy" == a rottmann <address@hidden> writes:
>
>     andy> Neil Jerram <address@hidden> writes:
>     >> Ken Raeburn, OTOH, is working on Guilifying the Emacs C
>     >> codebase, which would provide those primitives to Guile.  Guile CVS
>     >> also has several Emacs libraries that aim to make working with Guile
>     >> in Emacs easier.
>     >> 
>     andy> Hmm, is there some more up to date info available on this
>     andy> project?
>
> Which one - Ken's?  The last status I'm aware of was on this list
> several months ago - should be findable in the archive.  For further
> news I recommend emailing Ken.

I must apologize -- I've been stalled on this for at least a year.  I
got swamped at work, and this isn't exactly an easy project to work on
in a few spare minutes, so I didn't get back to it for a while, and
then a while longer, and so on.  Other, smaller projects started to
take up some of my time.  But I've been wanting for a while now to put
together the time to start up on this again.

My last work -- and soon, I hope, continuing -- was on the Emacs code
base, banging on various low-level macros and such to try to reduce
the dependence of the bulk of the Emacs code on the details of the
Lisp object implementations.  (For example, don't fetch the struct
window pointer from an object before checking that it is in fact a
window object, even though it "works" in the current implementation.
Don't make so many assumptions about which Lisp object references can
be treated as lvalues and assigned to.  Pass around Lisp objects and
get the struct pointers out of them, rather than assuming you can
trivially reconstruct the Lisp object handle from the struct pointer.)
That was the big, fairly pervasive thing I found I needed to do when I
first got a proof-of-concept hacked up version up and limping along;
to make the changes acceptable for inclusion in Emacs, they're being
redone more cleanly, and in smaller pieces.  Some string macros here,
now some lvalue/rvalue hacks over there, etc.  It's got a ways to go.
I forget, I think I was looking at the interval handling most
recently.

There was a bit of push-back from some of the Emacs developers, too.
Some "Guile compatibility might be nice, but no way are we going to
even consider replacing the working Lisp engine" (something I'm merely
hoping to defer any decisions or arguments on for a while), some "what
a waste of effort, nobody cares about Guile or Scheme" type attitudes,
"don't fix what ain't broke", etc.  Others, including RMS, do seem to
want at least the Guile compatibility.  (Well, unless things have
changed in the past year or so.)

I need to spend a little time getting back up to speed on the state of
things in the Emacs development world, but I'm hoping to pick up this
work again soon.  No reason other people can't help out with it too,
though....

Ken