On Monday, July 15, 2013, Rob Clark <robdclark@gmail.com> wrote:
> On Mon, Jul 15, 2013 at 5:25 PM, Peter Robinson <pbrobinson@gmail.com> wrote:
>> On Thu, Jul 11, 2013 at 7:40 PM, Brendan Conoboy <blc@redhat.com> wrote:
>>> On 07/11/2013 10:41 AM, Bill Nottingham wrote:
>>>>
>>>> Kernel, glibc, all the core library stacks. And I would argue that yes,
>>>> this
>>>> *includes* libGL. So llvmpipe needs fixed, outside of any desktops.
>>>> Should
>>>> we define the core functionality better? Probably.
>>>
>>>
>>> I would argue that it does not include libGL because it's not a requirement
>>> for headless deployment scenarios.  Why would you argue for it?
>>
>> I would argue that it's nothing to do with headless scenarios but more
>> that the vast majority of ARM GPUs support GL-ES which is a
>> sub/different standard of desktop GL (sorry, I'm not a graphics
>> programming expert!) and the support for that in mesa and in general
>> is terrible. There was a proposal to refactor mesa and when I spoke to
>> ajax (I think, sorry ajax if it wasn't you) or someone it wasn't
>> basically moving forward upstream at the moment. I'm not sure who
>> originally was driving this (my google fu doesn't give me the mailing
>> list proposal ATM).
>
> It is getting a bit off the topic, but this it isn't really a problem
> with mesa.  But rather that we have non-gallium closed src drivers
> from the GPU vendors in the ARM space, which only support GLES.  And
> most/all of the desktop stuff packaged in fedora (in particular,
> gnome-shell) is requiring GL.

This is incorrect. It uses cogl which has a gles renderer wich is supposed Tod work. If it does not that is just a bug that we should fix.