Bug 49020 – Need to simulate DEPTH_STENCIL in framebufferRenderbuffer / getFramebufferAttachmentParameter

• Component: WebGL
• Status: RESOLVED FIXED

[Zhenyao Mo]

Currently if DEPTH_STENCIL_ATTACHMENT is used, we attach to DEPTH_ATTACHMENT and STENCIL_ATTACHMENT in chromium --in-process-webgl and safari port.  We need to move this simulation to WebGLRenderingContext level so all ports including chromium command buffer will be covered.

Also, generate INVALID_OPERATION in getFramebufferAttachmentParameter if attachments conflict is detected (otherwise, for example, if query DEPTH_ATTACHMENT, DEPTH_STENCIL_ATTACHMENT might be turned).

[Zhenyao Mo]

Created attachment 73017 [details]
patch

Per discussion with Gregg, we do the enum mapping at WebGLRenderingContext level instead of in command buffer (and other lower-level ports).

Tested in Chrome command buffer, Chrome in-process-webgl, and Safari, all three ports.

[Kenneth Russell]

Comment on attachment 73017 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=73017&action=review

Generally looks good. It would be nicer to reduce duplication in the re-attachment code but I can't think of a more terse way to express it. One issue requiring a revision.

> WebCore/html/canvas/WebGLRenderingContext.cpp:2294
> +                                           0x88F0, // GL_DEPTH24_STENCIL8

This enum should be added to Extensions3D.h and the documentation for that class updated to indicate that it supports GL_OES_packed_depth_stencil and GL_EXT_packed_depth_stencil.

[Zhenyao Mo]

Created attachment 73021 [details]
revised patch: responding to kbr's review

[Zhenyao Mo]

Created attachment 73023 [details]
revised patch: updated the change log

[Kenneth Russell]

Comment on attachment 73023 [details]
revised patch: updated the change log

Great. Thanks for taking care of this.

[Zhenyao Mo]

Committed r71449: <http://trac.webkit.org/changeset/71449>