STR: http://people.mozilla.org/~bgirard/GfxInternDemo/webgl2demo/ The following demo works for me on the release channel.

I can't have a look before Thursday. If you are ready to wait a little bit, I can fix it. =)

Found the regression patch, and the winner is : bug 843668 But I can't edit the 'Deponds on' field of this bug :/

The frambuffer is incomplete right here : http://dxr.mozilla.org/mozilla-central/source/content/canvas/src/WebGLFramebuffer.cpp#144

(In reply to Guillaume Abadie from comment #2) > Found the regression patch, and the winner is : bug 843668 > > But I can't edit the 'Deponds on' field of this bug :/ Looks like you lost the editbugs privilege. They are a bit overzealous in clearing accounts at the end of internships. Let me look into it. Thanks for the help.

(In reply to Benoit Girard (:BenWa) from comment #4) > (In reply to Guillaume Abadie from comment #2) > > Found the regression patch, and the winner is : bug 843668 > > > > But I can't edit the 'Deponds on' field of this bug :/ > > Looks like you lost the editbugs privilege. They are a bit overzealous in > clearing accounts at the end of internships. Let me look into it. > > Thanks for the help. Thanks Benoit! Since bug 843668, there is a huge mix up between texture the internal format and the texture format, because of the renaming of ImageInfo::mFormat to ImageInfo::mInternalFormat, also replacing ImageInfo::Format() by ImageInfo::InternalFormat() everywhere, creating no sense line like : GLenum format = mTexturePtr->ImageInfoAt(mTexImageTarget, mTexImageLevel).InternalFormat(); in the code which is the main cause of this bug, And I think that should have created other regression even in WebGL 1 extensions, like this one in the depth texture extension shown here in WebGL 2, by rendering in a depth texture instead of depth render buffer. My mind is that the long-therm solution would be to add functions in GLContext to factorizes conversion between internal format to format and its backward transformation (like InternalFormatForFormatAndType in WebGLContextGL.cpp) at the very same place somewhere in a cpp file, and then completely forbidden any pattern like 'format = [whatever].InternalFormat()' by bringing back ImageInfo::Format() converting from ImageInfo::mInternalFormat. What do you think about that ?

There is some work going on in Bug 948002. Mea Culpa. It probably would have caused less issues to keep ImageInfo.Format() and extend with ImageInfo.InternalFormat() instead of renaming and falling over all subtle bugs that the change introduced.

I fixed the cause of the issue reported in Comment 3 in Bug 948002.