kWebGraphicsContext3DPerferredGLExtensions -> kWebGLPreferredGLExtensions

kWebGraphicsContext3DPerferredGLExtensions -> kWebGLPreferredGLExtensions

Also added a comment. This string should only be passed in for
WebGL. If it is being passed in for any other reason it is a bug
and needs to be fixed.

TEST=none
BUG=none

R=jamesr@chromium.org

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=86671

[greggman, 2011-05-24 22:02:51]

I know this isn't enough to prevent people from using this incorrectly but it's
a start.

[jamesr, 2011-05-24 22:05:10]

What string is the compositor supposed to be passing?

http://codereview.chromium.org/7067034/diff/1/content/renderer/webgraphicscon...
File content/renderer/webgraphicscontext3d_command_buffer_impl.cc (right):

http://codereview.chromium.org/7067034/diff/1/content/renderer/webgraphicscon...
content/renderer/webgraphicscontext3d_command_buffer_impl.cc:114:
kWebGLPreferredGLExtensions,
i think this codepath (render_directly_to_web_view) is only used for compositor
contexts, so this seems to contradict the comment above. should we be passing a
different string in?

http://codereview.chromium.org/7067034/diff/1/content/renderer/webgraphicscon...
content/renderer/webgraphicscontext3d_command_buffer_impl.cc:144:
kWebGLPreferredGLExtensions,
this is used for webgl, canvas 2d, and probably other stuff too :(

[greggman, 2011-05-24 22:24:15]

Everything else should be passing "*"

Is there is a way to distinguish the use case? Should I add an argument to
Initialize "isWebGL"? Or maybe add "isWebGL" to
WebGraphicsContext3D::Attributes?

[jamesr, 2011-05-25 01:38:48]

LGTM since this isn't changing anything.

Passing an attribute sounds like a good approach to me, but I'd see what kbr
thinks.  In the current code the case where render_directly_to_web_view==true is
never used for WebGL contexts, so you could try passing "*" for that path
without any other changes.