Re: sawfish-config segfault

[Teika Kazura <teika lavabit com>]

On Thu, 7 Jan 2010 13:33:03 +0100, Janek Kozicki wrote:
> Christopher Roy Bratusek said:     (by the date of Thu, 7 Jan 2010 13:24:00 +0100)
>> Valgrind gives nothing for sawfish-config, coredump only this:
>> 
>> #0  0x00007fe728ee4478 in ?? ()
> that's seriously weird.

Hey guys, glad the crash is fixed, but the sawfish-config is a shell
script, so you've got to embed debugger call into it. Have a look at
lisp/sawfish/cfg/main.jl.in:
------------------------------------------------------------------------
#!/bin/sh

# I tried: `REP_LOAD_PATH=... exec ...' but solaris /bin/sh won't play along

REP_LOAD_PATH="__nokogiridir__"
export REP_LOAD_PATH

exec rep "$0" "$@"

!#

;;; lisp code follows...
------------------------------------------------------------------------

Currently sawfish-config / -script are byte-compiled and put under
/usr/bin. In fact, I've been thinking that the bare sources, not
the byte-compiled ones, are better, because users can read them.
If they want a hack, then they can get it by editing. (I needed it
when I tried to solve the mystery of non-English letters
disappearance.)

Cheers,
Teika (Teika kazura)