On 18 Jan 2003, G. Roderick Singleton wrote:
>> I forgot to also mention, that if you check your Red Hat Linux
>> box, you'll find that it lists:
>>
>> Intel i740, i810, i815 and i830 only, and not i845.
>>
>
>Let me deal with the compatibility list issue. What I found was the
>lists were not available. As in can't look to see. Now you mention that
>RH or, for that matter, any other linux vendor can test all systems.
^^^
Can't. And that means can't test even 1/1000th the systems out
there. Not only are there hundreds if not thousands of video
cards in use. There are thousands of motherboards out there,
CPUs, BIOS revisions, etc... it is just impossible.
We do testing on some stuff we have, and we provide beta releases
to get user feedback. Of course all of this only applies to
hardware that contains XFree86 source code to begin with too.
>True as far as that goes. I see that as the purview of the
>manufacturers. That said, I see no reason for RH et cetera to
>make the existing list unavailable nor do I see any reason that
>they can't echo what they print on the boxen into these lists as
>a caveat.
Make what list? There is no list. I'm asked once a release what
video should go on the box, and presented with a list basically
of hardware that works, and hardware that is wanted to know if it
works. I generally remove all things from the list that are
shaky, and I make sure nothing is listed as supported if it is
totally unsupported.
I don't however have the time (nor does anyone here) to go
through each driver and figure out what all cards it supports in
code 'theoretically', and then test every single card in every
single resolution, with and without DRI, with and without video,
etc.. It's just amazing amount of work, work that should be done
by the video card companies.
People at Red Hat voluntarily test what hardware they own and
report bugs. Our QA department tests stuff and reports bugs.
Our beta testers do so also. Whatever isn't in bugzilla as a bug
is basically one of 2 unknowns:
1) Works
2) Doesn't work, we've no idea
So, there isn't really a heck of a lot of point to sending out a
random email here of what is and isn't supported IMHO. It is
just the way it is, and it is that way in all Linux/BSD OSs, this
isn't a Red Hat specific thing. Changing this situation would
require an amazing amount of work, even for just a little useful
info, and then even that isn't perfect.
What happens if I say "Radeon works great", and you buy a Radeon
67000 and it doesn't work? I do tell people Radeon is the most
supported hardware. Then they go out and buy a Radeon PCI card
and 3D acceleration doesn't work. ARGH. Unsupported. How do I
tell someone that Radeon A B C D and E are supported, but the PCI
version of B and C don't have 3D support, and Xvideo doesn't work
on E, plus DVI output doesn't work on A?
It is too complex to database all this info, plus, one single
patch coming in from the net, could invalidate all this data.
People asked "is i830" supported before. Answer was "yes". Then
they bought a Dell C400, and found out they only had 1Mb of video
RAM. Then it is "What is Red Hat doing to fix this? You said
i830 was supported!" How do I say i830 is supported, and at the
same time exclude the various random cases that it doesn't work?
It isn't XFree86 bug afterall, but it is a BIOS bug. These
oddities are there in every driver. It is just not possible to
say without a reasonable doubt "card foo will work 100%
guaranteed", or even "10% guaranteed".
The best and only way to know if something works, and yes this
sucks, is to look at the status doc on XFree86.org for the
release of X included in Red Hat Linux, choose a card listed as
supported, then ask on mailing lists about it, query bugzilla for
both open and closed bug reports, and query google for the same.
It sucks, but that is the way that choosing a video card in Linux
works right now. In order for us to make a serious claim that
something works 100%, we would then have to fix it when it
doesn't. While we do contribute fixes to X, it is not our
responsibility to maintain 20 video drivers on our own, and fix
all bugs in all of them. We do what we can, but it is ultimately
the responsibility of the video vendors who want to sell hardware
to Linux customers, and to the Linux community (of which we're
part of) to help do so also.
>Would have saved me two days of futzing around. As it is I bought a
>supported framebuffer and am happy.
Indeed, but this isn't easy problem to solve.
>Since emailing you, I have looked at the programmer guides for
>the 845 and 815 and see the vast differences and understand why
>things are slow. I also saw that Intel has an 850 and 870 in
>production so I guess this saga must continue forever.
Ah, so they do have them published now, cool. Another note for
the record.... I have only an Intel i740 video card. I have no
other Intel video hardware available. As such, i8x0 problems are
more or less "community supported best effort". Instead, I spend
most of my working time troubleshooting problems in the hardware
I do physically have, and have docs for. Mostly ATI, Matrox,
some 3Dfx, and various others.
Video hardware vendors and/or computer systems vendors wanting to
supply me with video hardware and/or systems with integrated
video that I do not have available for testing right now, can
contact me if they like off-list to arrange sending me hardware
to work with. In particular if docs are available publically, or
an NDA can be arranged then I can probably more actively support
other hardware too.
>I am willing to try helping if you would care to use my limited
>skills.
XFree86 project can always use more help indeed. ;o) Feel free
to join the devel@xfree86.org mailing list to help get started if
you like. If there are pointers I can provide you with, I'd
prefer to do it publically in hopes that it might reach the most
number of other potential developers as well.
Also, sorry if my email sounds a bit rantish... not intended that
way. I've gotten haggled about i830/i845 a lot lately and just
need to express myself and get it off my chest. Also, in hopes
someone else having problems will find this in the list archives
and not send me nastygrams. ;o)
Anyhow, thanks for the followup, and hope to see you on the
devel@xfree86.org list.
Take care,
TTYL
--
Mike A. Harris ftp://people.redhat.com/mharris
OS Systems Engineer - XFree86 maintainer - Red Hat
_______________________________________________
xfree86-list mailing list
xfree86-list@redhat.com
https://listman.redhat.com/mailman/listinfo/xfree86-list
IRC: #xfree86 on irc.redhat.com
