X Devs Drop NVIDIA Auto-Config Support

[Coppertop]

NV trying to forcefeed their stupid driver ... Honestly, this does it for me. I'll buy another NVidia card when pigs fly.

Who tries to force you to what? Read things twice before you write something like that. The nvidia's patch would make your X use nvidia binary driver ONLY IF you have it installed. If you don't - it won't download and install it for you, it won't stop your X from working or anything like that. Is it really so hard to understand? Or do I miss something here? Maybe some of you install drivers just for fun, and not in order to use them?

[Vadi]

Why would you have their driver installed if you're against it?

You're making no sense.

[deanjo]

since Git users probably don't want to use nvidia because of the ABI changes (as was mentioned in the mailinglist-discussion).

As also mentioned in the mailing list

This is not the problem you're making it out to be. If the server ABI
isn't supported, loading the driver will fail and the server will fall
back automatically.

[Crunchy]

Some people are saying "nVidia's patch", I wasn't aware it was submitted by an nVidia dev? I'm assuming it was an X.org dev that wrote the patch but others forced it to be removed.

As some others have pointed out it would only default to nVidia driver if it was installed. This makes sense to me as the open source versions are not usable if you require 3D but it probably can be left up to the distro to sort out.

[deanjo]

It also looks like the distro's are getting behind it. Stefan Dirsch from opensuse already replied.

On Sun, Jul 20, 2008 at 04:43:15PM -0700, Aaron Plattner wrote:
> Driver autoconfiguration doesn't impose anything on anybody. It simply
> tries to choose the correct default for the most people.

... which in this case makes perfect sense to me. At the time you've
installed the NVIDIA driver you've messed up your system anyway by
replacing GLX extension and libGL at least. So there's no use in
trying to use nv driver any more. But as usual this is all about
politics, and not about chosing the most appropriate driver the user
would expect to get.

Thanks for the patch, Aaron. I appreciate it. I assume most Linux
distributions will apply it. Silently, probably.

Best regards,
Stefan

[Vadi]

Some people are saying "nVidia's patch", I wasn't aware it was submitted by an nVidia dev? I'm assuming it was an X.org dev that wrote the patch but others forced it to be removed.

No, nVidia does contribute directly to the X.org and that was their patch. Which is actually quite good for the consumers if you, um, reason logically.

[Kano]

Well I would not overrate this patch. It is ok, when the X server would choose nvidia, but as it needs the nvidia kernel module and it is not 100% sure that this is there the xserver would fail - nv would still show a picture. Also the nvidia installer can already write a suitable config file it is highly unlikely that somebody removes it. Thats too special case for me, not really important. I highly doubt that there are so many user questions for that rare case...

[Melcar]

It also looks like the distro's are getting behind it. Stefan Dirsch from opensuse already replied.

It's no surprise really. Most distros always look for ways to increase their user base. It's the kernel and X guys that are really anal about this sort of things. I for one don't really care, as long as preference is given to the open source stuff.

[stan]

X.org devs are right

The X.org developers are right to reject NVIDIA's unscrupulous tactics. When NVIDIA's blob crashes, as it often does, less knowledgeable users will think it's X.org's fault. The X.org developers are just trying to protect themselves and the users from this hassle. If the user has to modify the xorg.conf file and change the driver names by hand, they'll have a clue that they're using an unsupported blob.

[deanjo]

The X.org developers are right to reject NVIDIA's unscrupulous tactics. When NVIDIA's blob crashes, as it often does, less knowledgeable users will think it's X.org's fault. The X.org developers are just trying to protect themselves and the users from this hassle. If the user has to modify the xorg.conf file and change the driver names by hand, they'll have a clue that they're using an unsupported blob.

OK so your saying the fact that they still have to go out of their way and still download and install the driver is not enough already but they should also be forced to drop to cli, scratching their heads instead of still having a failsafe fallback to troubleshoot the issue? That makes no sense at all. The only ones doing unscrupulous tactics are the xorg devs. It's not like nvidia is dropping the blob in xorg nor is it automatically magically installing it. If it is there, it will use it, again the user had to download and install it on their own. If the binary fails or does not see the nvidia blob then it reverts to nv. If anything it will lower the bug reports because it would eliminate people posting "I have no display, something is broken."