Re: I resign as Guile maintainer

[jldasch at 3-cities dot com (John L. Daschbach)]

As another scientist outsider like Valentin, though with even less
experience in guile, I would like to voice my concurence with his view. 
Guile does seem to be very usable as an extension language now, (modulo
the startup issues I have posted here recently).  I don't find it as
stable as SCM (for instance the CVS version I'm using will often hang in
the repl while loading a scheme file for the n'th time [e.g. I use a
scheme file to run some simulations, changing values in the scheme code
and 'load' ing it.  This will often hang after some number of repeats, but
killing and restarting the repl process works fine]).

There are number of issues that should probably be addressed for the guile
project to move forward.  Here is my list.

1.)  Documentation.  Enough has been said.  This is a real show stopper
for guile.  Much can be learned about guile by reading the scm
documentation, but with time this is less useful.  This must be resolved. 


2.)  Mission.  Many claims have been made for guile for years that seem
pretty bold.  The translation of other languages into guile, the rewriting
of emacs in guile, ...  This needs to be rethought.  Guile is, I think, a
great wrapper for C code, i.e. an extension language for C.  Others view
guile differently,  as something more akin to the next great language.  I
think this is perhaps a bit hopeful.  Scheme and Lisp will, I think, only
ever appeal to a select set of programmers.  I would wager that a large
percentage of VB programmers couldn't make heads or tails of scheme.  As
guile gets bigger, as more stuff moves into modules, it gets harder to use
as an extension language.  If the core of guile was R4/5RS scheme, plus a
few system extensions, and the C interface was frozen, and all this was
documented, I think it would help a great deal.  If someone writes a large
program and allows guile scripting it would be nice to know it will still
run in n years when compiled with the current version of the guile-core.

3.)  Publicity.  Somehow the guile scriptable projects get little mention
anywhere on the web I travel.  Not that I don't know about them, I just
don't often see reference to an application being guile scriptable.  You
do see this often for tcl, python, and perl.  Can this be changed?

I would be willing to contribute to a documentation effort.  There are a
couple limitations to this.  One, I don't know guile that well.  Two, I
have to get permission from my employer *even* though all work would be
done at home.  (I work for a U.S. National Lab, run by a contractor, and
all intellectual property is property of my employer unless waived, *even*
if it is non-commercial type work from home!).  I have written and edited
scientific works and could certainly contribute in these areas, as well as
possibly write or rewrite some areas of the manual, subject to peer
review.  Perhaps a few of us could generate enough energy to get a new
manual to production.

-John


-- 
-----------------------------------------------------------
jldasch@3-cities.com  (John L. Daschbach)
-----------------------------------------------------------