Customers have requested the OES_depth_texture WebGL extension be exposed in the near term for some deliverables for upcoming conferences. It is in draft status in the WebGL extension registry; for the time being it should be exposed via getExtension("WEBKIT_OES_depth_texture").
Sounds good. In general is there any process for how stable an extension should be before we land it in trunk? Since it's prefixed it is protected to some extent. (BTW - I have no opinion either way. I tend to think it's better to land features soon because you get more feedback that way)
(In reply to comment #1) > Sounds good. In general is there any process for how stable an extension should be before we land it in trunk? Since it's prefixed it is protected to some extent. An extension should have Draft status in the WebGL extension registry before implementation. Collectively the WG and community have agreed that extension "proposals" shouldn't be implemented, not even under vendor prefixes.
Gregg, did you mention you might have time to do this?
Me? I suppose I can but we don't have OES_depth_texture. We have ANGLE_depth_texture which has different restrictions. No cubemaps No uploading data (TexImage2D must be passed NULL, TexSubImage2D always generats INVALID_OP) No copying (CopyTexImage2D, CopySubTexImage2D) always generate INVALID_OP I don't think that should be exposed as OES_depth_texture since it's not OES_depth_texture
Agreed about the naming; in the process of renaming the extension to WEBGL_depth_texture and imposing the ANGLE limitations. Gregg indicated he's working on this, so marking Assigned to prevent accidental duplicated work.
Created attachment 148128 [details] Patch
Note: This page does not support DEPTH_STENCIL formats for depth textures. I'll add that in another patch.
Created attachment 148154 [details] Patch
Created attachment 148156 [details] Patch
Comment on attachment 148156 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=148156&action=review Looks great overall. Nice test. Two minor questions. Marking r+, but not touching cq bit in case you want or need to address either. > Source/WebCore/html/canvas/WebGLRenderingContext.cpp:3560 > + border, format, type, m_unpackAlignment); A little subtle given the offline discussions about ANGLE_depth_texture and the fact that it can't upload data to the texture. Is it worth a comment here that on some platforms, it may be tricky to implement initialization of depth or depth/stencil textures? > LayoutTests/fast/canvas/webgl/webgl-depth-texture.html:1 > +<!DOCTYPE html> Are there a few garbage characters here?
Created attachment 148223 [details] Patch
Comment on attachment 148223 [details] Patch Clearing flags on attachment: 148223 Committed r120668: <http://trac.webkit.org/changeset/120668>
All reviewed patches have been landed. Closing bug.