[Mesa-dev] Upcoming Mesa releases

[Ian Romanick]

On 05/18/2012 08:06 AM, Brian Paul wrote:
> On 05/18/2012 12:51 AM, Ian Romanick wrote:
>> On 05/06/2012 01:12 PM, Kurt Roeckx wrote:
>>> On Sat, May 05, 2012 at 09:18:24PM -0700, Ian Romanick wrote:
>>>> On 05/04/2012 03:57 PM, Kurt Roeckx wrote:
>>>>> On Fri, May 04, 2012 at 08:32:11AM -0700, Ian Romanick wrote:
>>>>>>
>>>>>> Part of the reason there are so few changes is that people get lazy
>>>>>> about cherry picking fixes over. :) Having regularly planned
>>>>>> releases (with reminder messages) should help keep people on track.
>>>>>
>>>>> I'm still missing patches marked as "candidate for stable
>>>>> branches" that where commited before the 8.0 release.
>>>>
>>>> I can believe that. Do you have a list?
>>>
>>> So I went over all the changes since on master since the branch,
>>> and put everything that was marked as candicate and things that
>>> looked like bugfixes to me on a branch. You can find it at
>>> git://github.com/kroeckx/mesa.git in the 8.0 branch.
>>>
>>> I've currently only tested that on sandy bridge, the piglet
>>> changes compared to the 8.0.2 release look like:
>>> crash->pass: 2
>>> fail->pass: 13
>>> skip->fail: 7
>>> skip->pass: 10
>>> pass->skip: 2
>>>
>>> From 8638/8777 to 8661/8792
>>
>> I've picked a bunch of these over. I didn't take ae92180 because it
>> enables an extension (GL_ARB_draw_instanced) that wasn't previously
>> enabled on 8.0.x. I'd do some more tomorrow.
>
> Here's a few that I'd like included:
>
> 1bc16bf98a1b5a4cca0c0ae2a80ba7982c6e4651
> f2724905469bbcde820b9ece2e4f44a610a75a65
> 99ed0b2f9b95a484aa78aeee0277bb669611be5b

Okay.  I'll pick those over.

>> Note: I haven't picked over any to core Gallium or Gallium drivers. I
>> can't test those, so I don't feel comfortable doing that on the eve of
>> a "stable" release.
>
> As we've said before, it's pretty easy to build/test softpipe at least.
> In any case, I didn't see any essential picks for llvmpipe, softpipe or
> svga.

I already spent about 3 hours picking, building, testing on one driver. 
  I'm not super excited about increasing that time.  Besides, me doing 
all the work doesn't encourage people to periodically pick patches to 
the stable branch like they're supposed to. :)

Our stable branch process may need some additional fine tuning.