Fixed gl_FragCoord emulation for point sprites of size > 1 using VPOS...

Fixed gl_FragCoord emulation for point sprites of size > 1 using VPOS
semantic for x and y coordinates rather than previous per-vertex
emulation when running on Shader Model 3.

Verified fix in Chromium with test case from bug.

BUG=http://code.google.com/p/angleproject/issues/detail?id=71
TEST=test case from bug
TBR=nicolas (LGTM'd by dgkoch)


Committed: http://code.google.com/p/angleproject/source/detail?r=473

[kbr1, 2010-11-05 01:24:18]

Please review. Thanks.

[nicolas, 2010-11-05 12:32:44]

http://codereview.appspot.com/2921041/diff/1/src/libGLESv2/Program.cpp
File src/libGLESv2/Program.cpp (right):

http://codereview.appspot.com/2921041/diff/1/src/libGLESv2/Program.cpp#newcod...
src/libGLESv2/Program.cpp:1376: mPixelHLSL += "    float4 gl_FragCoord :
VPOS;\n";
I don't think this is entirely correct. The VPOS semantic is only well-defined
for the x and y components. OpenGL ES 2.0 expects z and 1/w in the upper
components, but that's not guaranteed by Direct3D 9.

To fix this you'd need to combine both approaches.

[kbr1, 2010-11-05 18:53:50]

On 2010/11/05 12:32:44, nicolas wrote:
> http://codereview.appspot.com/2921041/diff/1/src/libGLESv2/Program.cpp
> File src/libGLESv2/Program.cpp (right):
> 
>
http://codereview.appspot.com/2921041/diff/1/src/libGLESv2/Program.cpp#newcod...
> src/libGLESv2/Program.cpp:1376: mPixelHLSL += "    float4 gl_FragCoord :
> VPOS;\n";
> I don't think this is entirely correct. The VPOS semantic is only well-defined
> for the x and y components. OpenGL ES 2.0 expects z and 1/w in the upper
> components, but that's not guaranteed by Direct3D 9.
> 
> To fix this you'd need to combine both approaches.

Thanks for your review. Updated the fix taking this feedback into account.
Re-tested with test case.

[dgkoch, 2010-11-05 19:50:10]

LGTM.