direct rendering .. help please

[Michel Dänzer]

On Tue, 2005-03-15 at 09:51 -0500, Vladimir Dergachev wrote:
> >> apart from the no fbconfigs warning), but I couldn't find a way how
> >> you'd output a more sane error message.
> >> If you disable it by default, noone will test or use it. Is it really
> >> important the xorg dri driver works by default (in cvs)?
> >
> > I don't think so.  cvs HEAD is the development branch and the latest
> > mesa will be separate/merged in for the next release anyway.  I say
> > leave tiling enabled.  If you don't know what you are doing use 6.8.x.

I'd spin this the other way around: People who know what they're doing
will get the 3D driver from Mesa and enable colour tiling anyway. Some
people obviously build X.org HEAD and expect it to work out of the box
though, and why shouldn't it? If the 3D drivers in there aren't
supported anymore, we should remove them.

> I generally agree with this, except that the point about very cryptic 
> failure modes is so true.
> 
> This needs to be fixed in Mesa drivers - if there is a mismatch or an 
> error of some sort it should be reported.

Even if it's fixed in the Mesa tree, that doesn't help the X.org tree.

> For example, why does it take LIBGL_DEBUG=verbose (and not even 
> LIBGL_DEBUG=1) to diagnose the issue of absent *_dri.so library or wrong 
> permissions on /dev/dri/cardX ?

Not sure, because anybody who doesn't need direct rendering in the first
place wouldn't want to be bothered with such messages?


-- 
Earthling Michel Dänzer      |     Debian (powerpc), X and DRI developer
Libre software enthusiast    |   http://svcs.affero.net/rm.php?r=daenzer