Youch! xemacs-beta memleaks PC to death when byte compiling
Stephen J. Turnbull
stephen at xemacs.org
Sun Feb 18 11:03:08 EST 2007
Asfand Yar Qazi writes:
> If I don't kill it early, it gobbles up all of the RAM, and I either
> have to reboot or wait for the kernel to kill it.
You make it sound like this happens *fast*. Again, this doesn't sound
like the leak reports I've heard before. I also just ran a full build
of the packages, including the haskell-mode that explodes for you; not
only did top report that XEmacs was using about 58M for the
haskell-mode compilation, but I never saw a figure higher than 75M.
(I'm not denying what you saw, just reporting that I don't see it.)
I gather this is your daily-use build, and you don't observe a
similarly disastrous leak in normal use, only in building the
packages? Is it always the haskell-mode.el compilation? (If not, you
might have a hardware issue, bad memory or something like that---not
likely, but something to keep in mind.)
To diagnose this, we really would like to have a full bug report; it
contains a lot of information about your build, including features
detected by configure as well as those you specifically requested. It
also reports on the Lisp environment (though that should not be
relevant in a package build).
To work around it, you could update to the latest CVS sources, as I
believe at least two leaks have been plugged since the release of
21.5.27. If you're consistently seeing the problem in haskell-mode,
you can just delete that whole directory, and any other that exhibits
the problem (except xemacs-base, debug, and ruby-modes, of course).
More information about the XEmacs-Beta