[Mesa-dev] [PATCH] scons: Add support for GLES.

[Chia-I Wu]

On Tue, Jan 25, 2011 at 10:08 PM, José Fonseca <jfonseca at vmware.com> wrote:
> Sorry for the delay. I thought I had replied but didn't.
> On Fri, 2011-01-21 at 01:04 -0800, Chia-I Wu wrote:
>> On Fri, Jan 21, 2011 at 12:18 AM, José Fonseca <jfonseca at vmware.com> wrote:
>> > Hi Olv,
>> >
>> > Looks good to me FWIW.
>> Thanks.
>> > I think we should avoid having opengl32.dll or the ICD loading
>> > glapi.dll, but that's not a reason to s given you've made it optional.
>> How to make scons build an ICD?
>
> The opengl32.dll serves double duty as an ICD. Instead of the wgl*
> entrypoints
>
> One can for example, replace the IHV's ICD driver with our opengl32.dll.
> It will work -- with pure software rendering only of course.
>
>>  I have to make opengl32.dll load
>> libglapi.dll, instead of statically linked, because _glapi_* are
>> declared __declspec(dllimport).
>
> Probably it shouldn't be declared dllimport when building the code in
> question...
It has to as libEGL.dll, libGLESv1_CM.dll, and libGLESv2.dll all use
_glapi_*.  Whoever defines those symbols has to dllexport.
>>  It might make some sense to not build
>> opengl32.dll when gles=yes is given.
> Perhaps. It's still a pitty we can't build everything simultaneously --
> it is easier to ignore issues when building just a subset.
I've committed the patch as is (urr, a while ago, sorry) as gles=yes
is optional.  So when the option is given, everything will get built,
but opengl32.dll will link to libglapi.dll.  There is still no
standard way to provide EGL/GLES on a desktop, so I think it will get
changed as there are needs or requests.  I read in one of the Khronos
slides that there is "EGL ICD" underway, but I don't know the
progress.

> Jose
>
>

-- 
olv at LunarG.com