On Fri, Nov 30, 2007 at 05:52:10PM -0800, Stefan O'Rear wrote:
On Fri, Nov 30, 2007 at 12:50:56PM -0500, David Roundy wrote:
On Fri, Nov 30, 2007 at 06:31:57PM +0100, Andrea Rossato wrote:
On Fri, Nov 30, 2007 at 07:46:41AM -0500, David Roundy wrote:
(Obviously) sounds like a memory leak. Presumably we're allocating X resources and never freeing them. But since they're unused, they get swapped out. I suppose this means carefully auditing all X calls. I wish we had *slightly* higher-level X calls that did GC for us (i.e. had finalizers to destroy objects when they're no longer in use). It'd be slightly problematic, since finalizers aren't guaranteed to be called at exit, so we might leak something when we restart, but that seems like a small danger relative to that of leaking while we run.
I just want to make it clear that I suffer from this problem well *before* switching to xmonad, so I don't think this is related to xmonad.
allbery_b once suggested it could be firefox, which is well know for allocating tons of pixmaps without freeing them...
Ah, I didn't catch this. That's a real bummer--it makes it far more challenging to catch bugs like this. :(
(Install and) run 'xrestop' to remove all doubt.
Hmmmm. For some reason xrestop shows xmonad as "<unknown>". Is there something we could to to make this remove more doubt? I can identify xmonad in the list by restarting xmonad, but that seems like a poor trick for catching leaks... -- David Roundy Department of Physics Oregon State University