stripping off "xf86-*-" from drivers

[Rémi Cardona]

Daniel Stone a écrit :
> On Sun, Jan 20, 2008 at 02:24:39PM +0100, Rémi Cardona wrote:
>> Lukas Hejtmanek a écrit :
>>> What makes you think that non modular drivers will be synced with X server
>>> changes? It is up to maintainers to watch X server changes and make patches as
>>> appropriate.
>> In Kernel Land (tm), if you modify a subsystem (say the USB one), you 
>> get to fix all the bundled drivers before Linus et al. accept your 
>> patch. Of course, that doesn't guarantee that the driver still works as 
>> intended, but that at least it builds.
> 
> Except for drivers which aren't built by default (a lot of them), which
> get compilation breakages.
> 
>> It's much easier to fix something that builds and runs.
> 
> No, not really.  Compilation errors are obvious.  If you just do a
> trivial substitution and stuff it up, then things get much harder.

Disclaimer, I'd rather the server and modules stay modular as they are now.

However, as others pointed out since I wrote this email, the issue at 
hand is social and not technical.

Maybe the Xorg world should adopt a model similar to Gnome or Mozilla ?

Both of them work more or less the same way with various freeze points 
before the final release. The major difference is that Mozilla has 
allowed itself to slip past imposed release dates to further improve QA. 
But overall, their system works fine.

Or like in Kernel Land (tm) where there's a 2 week merge window during 
which all hell breaks loose and then the rest of the time is spent 
ironing issues and making sure everything works.

Cheers :)

Rémi