On 10-Jun-2002 �yvind Stegard wrote:
> Hi,
> 
> Just posting a "little" issue I have with BlackBox, just to see if 
> anyone else is having similar problems.
> 
> It concerns version 0.65alpha8:
> After about 30-35 hours of operation (samme X session), BlackBox 
> suddenly dies for no obvious reason, and thus X exits => I lose unsaved 
> work etc etc etc. Is this some kind of memleak problem ? I am not 
> experienced with C++ so I wouldn't know, but something is going from 
> good -> bad in the course of those 30 hours, and in the end, Blackbox 
> chooses to end its life. I would like BlackBox to live happily ever 
> after (install). =).
> 

I suspect something else is happening.  You list Java, Mozilla and GAIM in your
application list -- all three of these seem to aggravate bugs in blackbox.

What you want to do is this:

first, build blackbox with debugging information for gdb.  When you run
configure use this command line as a base -- CXXFLAGS="-g -O2" ./configure. 
Add any --prefix or other flags you need.

to enable cores run this before you login -- 'ulimit -c unlimited'.  It is VERY
important that this happens before blackbox is started.

try to save blackbox's output to a file, I try to add messages around known
crash points.

When you have a crash and a core is generated the useful information is
obtained with this command -- gdb /path/to/blackbox /path/to/core.  Once gdb is
running the command to give it is 'bt' which stands for back trace.  This will
tell us where you were in the code when a crash occured.

Also try to be observant of the window manager when it crashes.  What were you
doing right before the crash?  What was the last action you did?  Currently a
major source of headaches is switching workspaces while any one of the 3
applications you listed is active.  Unfortunately I do not run Java apps or
GAIM so I find debugging this a little difficult.

Brad and xOr are currently working on pieces of this puzzle.  We hope to have
it solved soon.

I can be fairly confident in saying that time is not the problem here it is
more likely known bugs that are getting triggered.

Reply via email to