Bug 226865 – WebGL2 demo doesn't work due to failing compilation to metal backend

• Component: WebGL
• Status: RESOLVED FIXED

[Oliver Hunt]

The demo is at

https://zynaps.com/content/photon/

Checking for the error told me to file a bug :D

[Kyle Piddington]

Thank you for filing! I'll have a look!

[Radar WebKit Bug Importer]

<rdar://problem/79211560>

[Kyle Piddington]

I suspect this was a duplicate of 
https://bugs.webkit.org/show_bug.cgi?id=226660, due to 'metal' being a pretty common struct member in most tracers. This appears to be working on current Top-of-Tree webkit, can you verify this on your end?

*** This bug has been marked as a duplicate of bug 226660 ***

[Kyle Piddington]

I apologize, I closed this early. Reopening...

[Kyle Piddington]

Created attachment 433189 [details]
Patch

[EWS Watchlist]

Note that there are important steps to take when updating ANGLE. See https://trac.webkit.org/wiki/UpdatingANGLE

[Kenneth Russell]

Comment on attachment 433189 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433189&action=review

> LayoutTests/fast/canvas/webgl/shader-with-struct-array.html:36
> +        description("Tests that program compiling/linking with a reserved keyword.");

Remove "that" to make this grammatically correct?

> LayoutTests/fast/canvas/webgl/shader-with-struct-array.html:43
> +            runShaderTest(gl, vShaderSource, fshaderSource, "no error for using reserved keyword in struct")

Could you explain here what the reserved keyword is? This comment, and the description, don't seem to match the code change.

[Kenneth Russell]

Note: I filed:
Add glsl3 regression test for arrays-of-structs bug in ANGLE's direct-to-Metal backend
https://github.com/KhronosGroup/WebGL/issues/3298

about porting this new test into the WebGL conformance suite.

Would be great if you could put up a pull request, but no worries if not.

[Dean Jackson]

Comment on attachment 433189 [details]
Patch

+1 to Ken's comments.

[Dean Jackson]

Comment on attachment 433189 [details]
Patch

Remember to provide the expected results for the test :)

[Kyle Piddington]

Created attachment 433237 [details]
Patch

[Kenneth Russell]

Comment on attachment 433237 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=433237&action=review

> LayoutTests/fast/canvas/webgl/shader-with-struct-array.html:36
> +        description("Tests that program compiling/linking with a reserved keyword.");

Still have grammatical question about the description from the first patch.

> LayoutTests/fast/canvas/webgl/shader-with-struct-array.html:43
> +            runShaderTest(gl, vShaderSource, fshaderSource, "no error for using reserved keyword in struct")

Still have question about how accurate this comment is, relative to the code change. It looks more like a bug in handling of arrays of structs, than using reserved keywords.

[Kyle Piddington]

Created attachment 433238 [details]
Patch

[Kyle Piddington]

One more run, I forgot to change the result text.

[Kyle Piddington]

Created attachment 433556 [details]
Patch

[EWS]

Committed r279968 (239711@main): <https://commits.webkit.org/239711@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 433556 [details].