Re: [STUMP] stumpwm 100% CPU usage and no modeline

[Ted Zlatanov]

On Fri, 23 May 2008 15:20:00 -0700 "Shawn Betts" <address@hidden> wrote: 

SB> On Wed, May 21, 2008 at 10:04 AM, Ted Zlatanov <address@hidden> wrote:
>> stumpwm is taking up 100% CPU on my machine.  What can I do to debug
>> this?  I'm running under SBCL, and my setup is very unremarkable,

SB> I would recommend starting X but not starting stumpwm. You can start
SB> an xterm instead to keep X running. Then from a console load sbcl,
SB> asdf, then asdf load stumpwm, then start stumpwm with
SB> (stumpwm::stupmwm ":0"). 

It took me a while (including Google searches) to find that this could
be done with a SBCL script, as in
http://cl-debian.alioth.debian.org/repository/lcapello/stumpwm/stumpwm.sbcl
(and other places).

That should probably be documented in the repo.  I didn't see it.

I also recompiled SBCL from source because, according to the wiki, that
can make a difference in performance.  stumpwm is running under the new
SBCL now, in a screen session with a large scrollback buffer, so I'll
know if it starts grinding again.

SB> Then switch to X, get stumpwm spinning, switch back to the console,
SB> hit C-c to drop to the debugger and type backtrace to get a
SB> backtrace. then send that backtrace here.

SB> If anyone knows a better way to get such a backtrace, I'd love to
SB> hear it.

I don't know SBCL or the other CL environments, but ease of debugging is
IMO one of the most important aspects of open-source so I hope someone
here can help.  I don't mind the work I listed above, but most users
won't be able to handle it.

btw, I attached with strace to a spinning stumpwm and saw a LOT of
SIGSEGV events.  I don't know if that's expected with SBCL or if it's a
real issue that was related to the problem.

Ted