[STUMP] Workaround for SBCL hanging in run-prog-collect-output

[Julian Stecklina]

Hello,

I have been experimenting with a patch that works around SBCL's broken
run-program. You can find it in my Git repository[1] in the test branch
(which includes Shawn's test branch from savannah) and attached to this
mail.

Beware that this is not a nice patch and is highly dependent on SBCL
internals. I have tested this with SBCL 1.0.21 for about a day and it
seems to work for me. At least stumpwm didn't hang since.

If you have problems with stumpwm hanging on SBCL, please try this and
report back.

I am still working on a general solution, but it seems that
run-program's semantics are hard to emulate without using signals
(evil!) or threads (not always available). That being said, I have an
implementation (not complete, but enough for stumpwm's purposes) of
run-program that uses threads. My guess is that this won't be included
in stock SBCL, because run-program should really not depend on
multithreading being available...

Footnotes: 
[1]  http://cthulhu.c3d2.de/~blitz/git/stumpwm.git

Regards,
-- 
Julian Stecklina

Well, take it from an old hand: the only reason it would be easier to
program in C is that you can't easily express complex problems in C,
so you don't. - Erik Naggum (in comp.lang.lisp)

run-prog-workaround.diff
Description: Workaround for SBCL's run-program weirdness