uniform* entry points must ignore the call if a null WebGLUniformLocation is passed

[Kenneth Russell]

Per a recent WebGL spec update motivated by Gregg Tavares, passing a null WebGLUniformLocation to the uniform* entry points should cause the call to be silently ignored rather than generating an OpenGL error. This is the behavior of the underlying OpenGL implementation when a -1 location is passed to glUniform*, and is motivated by application frameworks which look up a bunch of uniforms, where some might not exist in a particular program or might have been optimized away by the GLSL compiler.

[Zhenyao Mo]

Created attachment 58319 [details]
patch

[Kenneth Russell]

Comment on attachment 58319 [details]
patch

Looks good. One minor comment about one of the tests.

LayoutTests/fast/canvas/webgl/script-tests/uniform-location.js:23
 +  shouldBeUndefined("contextA.uniformMatrix4fv(0, false, mat)");
For correctness we should change this to "contextA.uniformMatrix4fv(null, false, mat)".

[Zhenyao Mo]

Created attachment 58424 [details]
revised patch

Changed 0 to null in the test.

[Kenneth Russell]

Comment on attachment 58424 [details]
revised patch

Looks good.

[Dimitri Glazkov (Google)]

Comment on attachment 58424 [details]
revised patch

Energize.

[WebKit Commit Bot]

Comment on attachment 58424 [details]
revised patch

Clearing flags on attachment: 58424

Committed r61018: <http://trac.webkit.org/changeset/61018>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.