[webkit-reviews] review requested: [Bug 36262] Add GraphicsContext3D abstraction to WebKit API : [Attachment 50992] Patch

[bugzilla-daemon at webkit.org]

Kenneth Russell <kbr at google.com> has asked  for review:
Bug 36262: Add GraphicsContext3D abstraction to WebKit API
https://bugs.webkit.org/show_bug.cgi?id=36262

Attachment 50992: Patch
https://bugs.webkit.org/attachment.cgi?id=50992&action=review

------- Additional Comments from Kenneth Russell <kbr at google.com>
ChangeLog message follows:

Added WebGraphicsContext3D to the WebKit API and refactored Chromium's
GraphicsContext3D implementation to use it. All of the OpenGL calls have been
moved out of WebKit/chromium/src/GraphicsContext3D.cpp and into the
WebGraphicsContext3D implementation. GraphicsContext3D is still responsible for
the transfer of rendered output from the WebGraphicsContext3D to the
HTMLCanvasElement.

The GraphicsContext3DInternal class, which is a data member of
GraphicsContext3D for the Chromium port, remains. It is possible to eliminate
this class and thereby one level of delegation, but this is being deferred.

The needed entry point for a Chrome implementation of WebGraphicsContext3D has
been added to WebKitClient, but it is not being called yet by
GraphicsContext3D. It will be once this patch lands and Chromium is rolled
forward to support this entry point.

This is a large patch, but the transformation is almost entirely mechanical and
there is no change in functionality. Nearly all of GraphicsContext3D and
GraphicsContext3DInternal has been moved to WebGraphicsContext3DDefaultImpl.
The only area where the splitting of logic is less than mechanical is
GraphicsContext3D::beginPaint() and its callees.