How about releasing an update of xorg-x11-drv-intel for Fedora 11

[Terry Barnaby]

On 10/09/2009 12:19 AM, Dave Airlie wrote:
> On Thu, 2009-10-08 at 09:37 -0700, Adam Williamson wrote:
>> On Thu, 2009-10-08 at 14:05 +0100, Terry Barnaby wrote:
>>> No, I don't what to force testing on anyone (although F11 has done
>>> that
>>> already :) )
>>> I was just suggesting that a separate yum archive with the packages
>>> necessary
>>> to test the later graphics development code that will be in F12 could
>>> be
>>> made available for people to try out easily with their F11 systems.
>>> They can optionally try these. I think it will allow 3D to work for
>>> many people
>>> (from my experience of the latest GIT versions) although others would
>>> not be so
>>> lucky. They can easily back these changes out if they have more issues
>>> than
>>> the standard graphics system.
>>
>> Then please feel free to make one. :) I don't mean that in a snide
>> fashion, but it really is the answer. As noted, having our X.org
>> developers spend time on such a repository directly subtracts that
>> amount of time from the time they would otherwise spend actually
>> developing the drivers (our X.org maintainers are also major upstream
>> developers) and fixing reported bugs.
>
> I thought about doing something like this the other day, but really
> if we had something like Ubuntu PPA, which I think is on the longterm
> plans for Fedora then it would be a lot easier to do.
>
> At the moment its just too distracting to do.
>
> The thing with doing updates for F11 is the regression rate due to
> lack of QA, I put Mesa packages into updates-testing that fixed a
> lot of r300/r500 bugs back at the start of F11 and it went into
> testing a few weeks later and broke Intel, I got 0 reports during that
> u-t phase about breakage. So now I have a package in stable that
> lets 3D works for x num of people and breaks compiz for y number.
>
> So I've pretty much given up on pushing anything to previous Fedora
> releases that isn't a security fix or major crash fix, because we simply
> don't have the QA in place to avoid regression current users, at least
> if you install F11 on your hw, and it doesn't work well, you know that,
> if you install it and it works well then later stops working well, thats
> a lot worse situation to end up in.
>
> I think for F12 updates we could really do with some sort of side repo
> setup, so we could have a stability period where QA could happen on
> packages that may end up in updates a month or two later.
>
> Dave.
>
>> Given that there is a lot of development work to do on these drivers
>> (which is why you find the newer versions better...) and a lot of bugs
>> to fix (we generally barely keep up with the rate of bugs filed as it
>> is), we don't see that as a good trade-off. You'd get backported drivers
>> for stable releases, but the rate of development of the actual upstream
>> drivers would be noticeably slowed, and fewer reported bugs would
>> ultimately get fixed.
>>
>> Backporting packages is not intrinsically very difficult, though it is
>> somewhat time-consuming, so it's something for which a far greater
>> candidate pool exists than X driver development. Thus, the suggestion
>> that someone else do it. For instance, you. It seems you've already
>> successfully built the latest versions of things locally; if you can do
>> that, you can put them in a package and put the package in a repository,
>> it's not a very hard process and it's all documented on the Wiki.
>>
>> --
>> Adam Williamson
>> Fedora QA Community Monkey
>> IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org
>> http://www.happyassassin.net
>>
>
>
I totally agree with you on the QA issue. Maybe I am wrong, but I haven't
seen any real set of tests to be performed on Fedora 3D graphics.
I tried the ATI test day for graphics. On the 3D graphics side it said to
run "glxgears" and if you like other 3D apps that you use. Running your
other 3D apps is difficult from a limited Live distribution...
I really think a simple test procedure should be implemented and documented
to at least check for basic functionality with the main 3D applications.
Ideally an automatic test program should be part of this.
This would allow a relative novice to test the 3D system on their hardware
and hopefully automatically feed back constructive results from the
myriad of different graphics hardware options.