[Xcb] xlib

[Jamey Sharp]

On Fri, Mar 04, 2005 at 08:28:15PM -0500, Jeremy Kolb wrote:
> Jamey Sharp wrote:
> >I assume you're referring to the occasional assertion failures from
> >instances of the _xcb_assert_valid_sequence macro that is declared in
> >xcbint.h and used by both XCB and Xlib.
> >
> >...
> 
> Thanks Jamey, how do I find out when this is happening?  I can't make 
> any sense of ethereal for the wire stuff, it's just too confusing. 

Temporal stuff is hard, and probably really important for this bug.
Maybe inserting some tracing printfs in _X{Put,Get}XCBBuffer and
XCBSendRequest would help?

Hmm. We could add a flag in the XCBConnection to note when Xlib owns the
buffers, so that XCB could assert() ownership whenever it was about to
touch the buffers. That might be helpful.

Running under valgrind would probably not be a bad idea either. I
haven't checked for uninitialized variables in XCB for a while.

> Also, gdb doesn't give me any function names on a backtrace even though 
> I load the shared-library files and have everything compiled with -g.

Erm... You may find it helps to turn off optimization, but I can't
promise that. What application are you testing with? Maybe there's a
simpler one that still shows the bug and is easier to work with.

--Jamey