Re: [Mesa3d-dev] Last minute additions for Mesa-5.1 release?

[Ian R. <id...@us...>]

Brian Paul wrote:
> Ian Romanick wrote:
> 
>> I have a couple additions that I'd like to commit before the Mesa-5.1 
>> release.  They're both pretty minor.  The first adds support for 
>> GL_IBM_multimode_draw_arrays.  The code for the two added functions is 
>> almost identical to the code for GL_EXT_multi_draw_arrays.  The other 
>> change adds support for GL_ARB_point_sprite.  This extension is a 
>> subset of the already supported GL_NV_point_sprite extension, so the 
>> changes are trivial.
> 
> OK.
> 
>> I have two questions.  First, is it too late to add new features for 
>> the 5.1
> 
> not at all.
> 
>> (or even the 5.0.2 in the case of ARB_point_sprite) release?
> 
> I'd rather limit 5.0.2 changes to bug fixes only.

Okay.

I guess the next (rhetorical) question is, "How soon after Mesa 5.1 is 
release can we merge that to DRI?" :)

>> If it's not, where should I commit the changes?  The new tree or the 
>> old tree?  I've been doing my work in the new tree, so I hope that's 
>> where it should go! :)
> 
> Check in your changes to the Mesa-newtree trunk.

Great.  Will do.

>> I have included the patch that adds IBM_multimode_draw_arrays.  This
>> patch also adds the extension string and non-static entrypoints for 
>> SUN_multi_draw_arrays.
> 
> Never heard of that one.  It's not in the OpenGL extension registry.
> Is it really identical to the EXT extension?

They are 100% identical.  From the EXT_multi_draw_arrays spec (which 
lists both extension strings):

     (Note: this extension has been promoted from SUN to EXT status.
     Implementations should advertise both name strings, and both EXT
     and SUN versions of the new GL functions should be provided).

>>  There are a few older apps that still just look for the Sun extension 
>> and at least one closed-source DRI driver exports it.  I also changed 
>> add_newer_entrypoints to be table driven.  This make context.o shrink 
>> by about 3KB.
> 
> Good.
> 
>> +# GL_IBM_multimode_draw_arrays (XXX unofficial dispatch offsets!!!!)
>> +
>> +name        MultiModeDrawArraysIBM
>> +return        void
>> +param        mode        const GLenum *
>> +param        first        const GLint *
>> +param        count        const GLsizei *
>> +param        primcount    GLsizei
>> +param        modestride    GLint
>> +category    GL_IBM_multimode_draw_arrays
>> +offset        708
>>  
>> +name        MultiModeDrawElementsIBM
>> +return        void
>> +param        mode        const GLenum *
>> +param        count        const GLsizei *
>> +param        type        GLenum
>> +param        indices        const GLvoid * const *
>> +param        primcount    GLsizei
>> +param        modestride    GLint
>> +category    GL_IBM_multimode_draw_arrays
>> +offset        709
> 
> Those two functions are already in APIspec, though incomplete.  Can you 
> fix that up?

Okay.  I see that the 'mode' parameter for MultiModeDrawArraysIBM is 
wrong (as it is in glext.h), so I'll fix that too.  Are those offsets 
okay?  I don't know what the policy is for assigning offsets to functions.