Bug#639737: fixed in nvidia-graphics-drivers 280.13.really.275.28-1

[Maurizio Oliveri]

> What is the subject of bug #639737?
>    "nvidia-glx does not support xorg-video-abi-11 required by
>     xserver-xorg-core 1.11.0"
> 
> *That* was fixed in 280.13.really.275.28-1.
That was 'fixed' with what looks like a temporary release with many
graphical issues, probably for those who rather have an upgraded xorg
able to actually render something with nvidia drivers for developing
reasons than a fully working xorg... Which makes other things, such as
gui testing, pretty hard...
The nvidia driver still doesn't support the new xorg abi (not
fully, although they claim so, most features aren't actually working),
that's the main reason of the new issues coming out.

> And *this* is a new bug, a regression NVIDIA introduced in their new
> release, so it deserves a new bug report. But there is already one:
> http://bugs.debian.org/641344 "Graphics errors after upgrading to
> xserver with abi 11" - is it the same you are experiencing?
>
> Anyway, NVIDIA needs to be told about these problems, please follow
> their bug reporting instructions:
> http://www.nvnews.net/vbulletin/showthread.php?t=46678
They probably already know about this:
http://www.nvnews.net/vbulletin/showthread.php?t=166025
The problem here is that as for that post on nvidia forum, they said it
could be a bug in xorg's abi. Yet it could actually be a feature. So
nvidia released a not-fully-working driver that notfully-supports
xorg's new abi... So what now? It's either a xorg bug for having broken
abi, or a nvidia bug for not being able to support the new abi...
Opening a bug for any of them would probably end in a mutual blame...

> Can't reproduce this here. X starts fine with this minimal xorg.conf:
> 
> Section "Device"
>         Identifier "GPU"
>         Driver "nvidia"
> EndSection
> Section "ServerFlags"
>         Option          "IgnoreABI" "false"
> EndSection
Mh.... I actually tried upgrading again today to run
nvidia-bug-report.sh, but I wasn't able to reproduce it either... Dunno
what to say here, yesterday I did several tries, and it only worked
after removing ignoreABI false... Got to apologize about that, probably
there was something else in the way

> What do you expect from using Debian unstable/sid? A well tested
> distribution? Sid *is* the testbed. Sometimes things break.
I know that. Yet it's useable thanks to stuff like apt-listbugs, and
I'm not really complaining about that. I just don't see why it's
considered better to include a not fully working (and older)
nvidia driver to fix this issue, when it just causes a good lack of
performance (and it's probably just a workaround until a real solution
is found)... I mean, it's new xorg vs new nvidia (since the 'new
driver' is actually a patched old release), woudln't it be better to
leave things as they were (or, at least, warning about the possible
performance lost when upgrading) until nvidia/xorg come to a really
working solution?

---Maurizio