Re: [Chromium-dev] OpenGL-ES

[James A. <ame...@vi...>]

>James Amendolagine wrote:
>> I believe that I've come full circle on this. I'm going to do my first
>> try at this as an extension, or an extension-like addition. 
>> 
>> Just for some feedback on the documentation, it looks like the extension

>> addition instructions are out-of-date. 
>> 
>>   * Update the extension strings in spu/state_tracker/state_limits.c as
>>     well. 
>>   * Extensions which add new OpenGL functions are more work. The OpenGL
>>     API dispatcher is defined by the functions in
>>     cr/glapi_parser/system_header/CR_GL.H. New API functions must be
>>     added there. 
>> 
>> 
>> 
>> 
>> I could not find spu/state_tracker/state_limits.c
>> or glapi_parser/system_header/CR_GL.H
>
>I'm updating the docs now.  state_limits.c is in the cr/state_tracker/ 
>directory.

I hope that I am not double posting here. It seems like my other e-mail is broken..


> Offhand this looks good. Some early examples of the es gl.h were missing
> the glGet commands (and chromium might like to have these available) but
> I believe that the spec says that they should be there. 
> 

I was wrong about this. The get's are delebirately missing in the ES 1.0spec.
(I was looking at the 1.1 spec). 



As an experiment, I've added one ES call to the system glClearColorx I did not
add the fixed datatype, just added a the call name. -- After adding the call
to Mesa to test, I was able to run an app with the call and it made it through
a modified app->chromium->modifiedMesa->gfx. 



I kept notes on every modification that I needed to do, so maybe at one point
I can give them to you to update the documentation.

 This is my current take on doing this port (I would like some feedback):I can
just add the entire GLES API as the standard path-gl, and whatever GLES extensions
as extensions. So Chromium will be a superset of GL and GLES. There may need
to be some modifications to the gl loading calls to attach the es libraries.


I have not throughly looked at this yet. I have also not looked at the glx/egl
fixups required. 

Assuming that the loading and glx/egl portion is smoothed out, this is the scenereo:
-Chromium will then be the superset of (gl+gles) 
-When the gl portion initializes, it will discover if the system is GLor GLES.

-If the system is GL, then application es calls can either error out, or be
converted on the fly to the appropiate GL call by the render spu. 
-If the system is GLES then the application GL calls can error out, or possibly
be converted on the fly like above. 


Jamie 







>
>CR_GL.H no longer exists.  New functions are defined in the 
>glapi_parser/APIspec.txt file.
>
>-Brian
>
>
>
>-------------------------------------------------------
>This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
>Project Admins to receive an Apple iPod Mini FREE for your judgement on
>who ports your project to Linux PPC the best. Sponsored by IBM.
>Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
>_______________________________________________
>Chromium-dev mailing list
>Chr...@li...
>https://lists.sourceforge.net/lists/listinfo/chromium-dev
>
>