On Mon, 2004-08-09 at 23:50, Pavel Cahyna wrote:
> A bit off-topic, but are you planning merging utah-glx to X.Org Xserver?
> Could it be more portable than DRI? Because AFAIK it does not require any
> special kernel support. Also, it can accelerate rendering done by clients on
> other machines (think about X-terminals). I had never tried it, but I plan
> to.
> 
> Bye	Pavel

Utah-glx is only able to manage doing DMA by doing nasty hacks, which
kills your portability and the ideal of needing less configuration.  I
(and the rest of the DRI developers, I bet) would much rather see people
working on porting drivers to the DRI infrastructure, and working on
getting accelerated indirect rendering working, than spending time on a
pure userland solution that at its best will rely on nasty hacks and not
be able to use kernel assistance for things that really need it.  The
project also seems pretty dead, given the mailing list.

It's not terribly hard for someone with some kernel knowledge to port
the DRM, I don't think.  I did quite a bit of it with essentially no
kernel or driver knowledge at all.  Plus, there are DRM maintainers for
Linux, and one for FreeBSD (me) that's happy to answer questions for
people porting.

Given that, no, I don't expect utah-glx to ever be merged to X.Org.  For
someone interested in jumping into 3D stuff, there's porting to other
*BSD, as well as many in-development drivers to work on (savage, mach64,
sis6326, trident, s3virge, tdlabs, what else?), along with testing and
improving conformance of existing drivers.

-- 
Eric Anholt                                eta@lclark.edu          
http://people.freebsd.org/~anholt/         anholt@FreeBSD.org