[compiz] compiz 0.7 error "No GLXFBConfig for default depth"

[Colin Guthrie]

Colin Guthrie wrote:
> Frederik Himpe wrote:
>> I am 100% sure it is caused by compiz update. I do a complete restart
>> of my computer every day. Let's take a look at the transactions done
>> around that time:
> 
> OK I'm convinced ;) Now just to work out why.....

Hi Compiz People :)

I've been trying to help Frederik and other nvidia users get to the
bottom of this. It seems from various tests that this is indeed caused
by compiz 0.7.0. Reverting to 0.6.2 and it works.

I created a debugging patch that allowed us to see where in the
initialisation things bail out.

The debug patch is here: https://qa.mandriva.com/attachment.cgi?id=9040
It basically just adds loads of info messages in.

The output on Frederik's setup is here:
https://qa.mandriva.com/attachment.cgi?id=9044 (LZMA compressed log file)

According to maniac103 whom I spoke to on #compiz-fusion-dev it fails at
"Continue point 7" which is mipmap

Feb 29 13:07:31 <maniac103>	@later tell coling The mipmapping seems to
be the difference ... for your fellow, it fails at "Continue point 7"
where it works here (in other words: when it fails at c7 for him, it
passes here; where it fails at other places for him, it also fails here)
Feb 29 13:08:02 <coling>	maniac103, Yo
Feb 29 13:08:19 <maniac103>	coling: but onestone might be more clueful
about this than me ;-)
Feb 29 13:09:19 <coling>	maniac103, cool. I presume you don't know why
it would fail? Other than slightly different h/w? THere was a new nv
driver out a couple days ago too... Not sure if he's tested with that
**** This has now been tried ****
Feb 29 13:09:43 <maniac103>	coling: no, I'm no expert in GL and GLX :-(


So is anyone able to shed more light on this confusing problem. I'm
curious as to if this is a Mandriva only issue but our compiz is fairly
unpatched (only really changing defaults and adding stuff from upstream
git).

I've tried to compare screen.c from 0.6.2 version to the 0.7.0 version
and that whole block seems not to have changed. So could there be
something else that would effect this in anyway?

Cheers for any insights.

Col