Bug 40496 – fast/canvas/webgl/get-active-test.html failed in Chrome Mac

• Component: WebGL
• Status: RESOLVED FIXED

[Zhenyao Mo]

--- /b/slave/webkit-rel-mac-webkit-org/build/src/webkit/Release/../../../layout-test-results/fast/canvas/webgl/get-active-test-expected.txt 
+++ /b/slave/webkit-rel-mac-webkit-org/build/src/webkit/Release/../../../layout-test-results/fast/canvas/webgl/get-active-test-actual.txt 
@@ -25,13 +25,13 @@
 PASS info[ii].type is expected[ii].type
 PASS info[ii].size is expected[ii].size
 PASS info2[ii] is non-null.
-PASS info2[ii] is non-null.
-PASS info2[ii].name is expected2[ii].name
-PASS info2[ii].type is expected2[ii].type
-PASS info2[ii].size is expected2[ii].size
-PASS info2[ii].name is expected2[ii].name
-PASS info2[ii].type is expected2[ii].type
-PASS info2[ii].size is expected2[ii].size
+FAIL info2[ii] should be non-null. Was null
+FAIL info2[ii].name should be ival. Threw exception TypeError: Cannot read property 'name' of null
+FAIL info2[ii].type should be 5124. Threw exception TypeError: Cannot read property 'type' of null
+FAIL info2[ii].size should be 1. Threw exception TypeError: Cannot read property 'size' of null
+FAIL info2[ii].name should be ival2[0]. Was u_modelViewProjMatrix.
+FAIL info2[ii].type should be 5124. Was 35676.
+FAIL info2[ii].size should be 2. Was 1.
 PASS context.getActiveAttrib(program, 2) is null
 PASS context.getError() is context.INVALID_VALUE
 PASS context.getActiveAttrib(program, -1) is null

[Zhenyao Mo]

Will roll out the failing part first, then investigate.

[Zhenyao Mo]

Manually rolled out the failing part of the test: http://trac.webkit.org/changeset/61038

Will add it back once this bug is fixed.

[Zhenyao Mo]

Created attachment 58838 [details]
patch

Please review this first.  Commit will wait until command-buffer port in Chrome is fixed.

[Kenneth Russell]

Comment on attachment 58838 [details]
patch

Argh. Thanks for fixing this.

Could you please fix the same bug in Chromium in src/chrome/renderer/webgraphicscontext3d_command_buffer_impl.cc ? Either that, or we should refactor the makeContextCurrent() calls into the GraphicsContext3DInternal class in WebKit/chromium/src/GraphicsContext3D.cpp so that ports don't need to remember to do this.

[Zhenyao Mo]

(In reply to comment #4)
> (From update of attachment 58838 [details])
> Argh. Thanks for fixing this.
> 
> Could you please fix the same bug in Chromium in src/chrome/renderer/webgraphicscontext3d_command_buffer_impl.cc ? Either that, or we should refactor the makeContextCurrent() calls into the GraphicsContext3DInternal class in WebKit/chromium/src/GraphicsContext3D.cpp so that ports don't need to remember to do this.

Yes, I'll fix the command_buffer port.

[Kenneth Russell]

(In reply to comment #5)
> (In reply to comment #4)
> > (From update of attachment 58838 [details] [details])
> > Argh. Thanks for fixing this.
> > 
> > Could you please fix the same bug in Chromium in src/chrome/renderer/webgraphicscontext3d_command_buffer_impl.cc ? Either that, or we should refactor the makeContextCurrent() calls into the GraphicsContext3DInternal class in WebKit/chromium/src/GraphicsContext3D.cpp so that ports don't need to remember to do this.
> 
> Yes, I'll fix the command_buffer port.

OK, thanks. This patch looks good, but we should consider refactoring where the makeCurrent() calls occur.

[Dimitri Glazkov (Google)]

Comment on attachment 58838 [details]
patch

ok.

[WebKit Commit Bot]

Comment on attachment 58838 [details]
patch

Clearing flags on attachment: 58838

Committed r61471: <http://trac.webkit.org/changeset/61471>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.