nvidia-glx

[Florian Zeitz]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Rocco Stanzione schrieb:
> On Monday 06 March 2006 05:03, Sam Morris wrote:
>> While you are basically correct, this is the wrong approach. The X
>> server should create a list of suitable drivers for the video card in
>> question when it starts, and try each one in turn until it finds one
>> that works. In this case it would be {nvidia, nv, vesa}.
>>
>> We've had a working hotplug system in Linux that basically does the same
>> thing, but for kernel modules, for years now. I had hoped that a similar
>> method would be adopted by the X server now that it's free of the
>> retarding grip of xfree86.org, but it hasn't yet. :(
> 
> This is exactly what I think the ideal solution is.  I don't know if we want 
> to be hacking around in Xorg on our own to implement it, but meanwhile I 
> think we could come up with a reasonable workaround.  We have a directory of 
> configuration files (xorg.conf.drivername) for installed 
> drivers, /etc/X11/xorg.conf is a symlink to /etc/alternatives/xorg.conf and 
> startx/gdm/kdm/etc. start X via a script that tries these configuration files 
> in order.  In addition to the other arguments mentioned in the thread, this 
> would allow me to start my machine directly into X, albeit without hardware 
> video acceleration, between installing a new kernel and building/installing 
> nvidia drivers for it, by failing over to the nv driver.
> 
> Rocco Stanzione
> 

I think it would take WAY to long if you'd actually do this. For me
starting an X session (no matter if it fails or not) takes roughly 5
seconds. Now imagine if you'd try all possible drivers which are really
a lot and maybe the 20th would work, that'd be a waiting time of a
hundred seconds. Also this way it would not be possible anymore to edit
your xorg.conf which you have to do sometimes (configuring special input
devices etc.).
Maybe this would work in combination with parsing lspci and having a
database with possible output -> xorg.conf mappings, but that'd defeat
the purpose.

Florob
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEDJZd0JXcdjR+9YQRAv9TAJ92i0mqnB0sp0S8/xSrzcWnro5nsACgizqG
8RYlKY/Ag9OB18PXbrzhNew=
=iZ3f
-----END PGP SIGNATURE-----