[Ecls-list] ECLM 2008

Juan Jose Garcia-Ripoll jjgarcia at users.sourceforge.net
Wed Apr 23 13:06:23 UTC 2008

On Wed, Apr 23, 2008 at 9:35 AM, Didier Verna <didier at lrde.epita.fr> wrote:
>   As both a core maintainer of XEmacs[1] and a Common Lisp advocate, I'm
>  very interested in your "Wild Idea #3" :-) Do you have more pointers
>  about that ?

Indeed. I worked on that long time ago, as one of the first ECL
projects. The first thing I did was to compare Emacs and XEmacs. I
found the latter to have a cleaner code base. Then I worked on
replacing XEmac's garbage collector with Boehm-Weiser's. This was a no
brainer. I mean, it requires work, but it _was_ pretty easy. The thing
built and I could bootstrap XEmacs under Windows (I think it was
Cygwin), which is what I was using at the time.

Things worked pretty well. I was getting certainly an overhead due to
the more imprecise garbage collection, but nothing beyond 50% the size
that ordinary XEmacs used. And I used none of the tricks to make B-W
more precise.

So my original  plan was to link ECL into XEmacs _without_ eliminating
elisp. Why would one? I mean, one can simply expose all datatypes from
XEmacs to ECL using the FFI. Also to provide a compatibility layer
with the functions that are accessible to elisp for manipulating
buffers, files, etc. One this works, it is time to think on moving
those XEmacs types to ECL ones. Like conses, buffers, etc, they can
all be turned into native ECL types.

>  Footnotes:
>  [1]  note how the first 2 letters of the name are upcased ;-)

Hehe :-) I have problems with computer program names. But the biggest
difficulties are of pronunciation. My god, CLOS, CLISP, ... I had real
problems preparing the talk



Facultad de Fisicas, Universidad Complutense,
Ciudad Universitaria s/n Madrid 28040 (Spain)

More information about the ecl-devel mailing list