[ FYI to those who may have posted something to the CTWM list
   to which I might be expected to respond: something seems to
   be wrong with email either at the server end or at my end (or
   somewhere in between) because I am not seeing my own posts
   to the list, even though I occcasionally see other people
   mentioning them.  I must therefore assume I'm intermittently
   failing to receive CTWM messages.  (I've already mentioned
   this privately to Richard) I will continue to post to the
   list in hopes that this is soon cleared up but please CC:
   me in any messages that you want to be sure I will see...  ]

>> I too have noticed that ctwm crashes for me about once a month
>> or so, but I could never figure out why.  So like you, I ran
>> ctwm in a loop ;-)
>
> Strange.  It never seems to crash for me, using Ctwm with Fedora
> 16 on both laptop (Dell E6410) and desktop, both with integrated
> intel graphics.
>
> I wonder if it could be an interaction with a specific application?

Quite possibly, if such interaction could get X into some
state where it's (temporarily?) unable to respond as expected,
which in any case is definitely what was leading CTWM to take
a code path that leaves the global Scr pointer variable NULL,
resulting in the SEGFAULT that was plaguing me.  That may or may
not be the same bug that [email protected] was enjoying -
there are so many moving parts that I can't provide a universally
applicable recipe for reproducing the fault.

Fortunately, once I saw that Scr was being left NULL when
either XFindContext() or FindScreenInfo() [which relies
on XGetWindowAttributes()] didn't answer as expected, it was
easy to add the missing defensive code, and I haven't seen the
problem since...
 

Reply via email to