[webkit-reviews] review requested: [Bug 46894] Redesign extension mechanism in GraphicsContext3D : [Attachment 72785] Patch

[bugzilla-daemon at webkit.org]

Kenneth Russell <kbr at google.com> has asked  for review:
Bug 46894: Redesign extension mechanism in GraphicsContext3D
https://bugs.webkit.org/show_bug.cgi?id=46894

Attachment 72785: Patch
https://bugs.webkit.org/attachment.cgi?id=72785&action=review

------- Additional Comments from Kenneth Russell <kbr at google.com>
>From the ChangeLog:

Upon request, factored out extension support from GraphicsContext3D into a new
Extensions3D class. (The plural was chosen because the class and subclasses
hold multiple extensions.)

Unlike GraphicsContext3D, Extensions3D contains only pure virtual methods. This
was done because Extensions3D's inheritance diagram and usage pattern is very
different from that of GraphicsContext3D, and the concrete subclasses need to
decide how to implement the various entry points. Requiring them to be placed
at the Extensions3D level will cause implementation details to leak into the
base class, which is highly undesirable. Any virtual call overhead to these
entry points will be negligible.

Changed call sites utilizing these extensions to call through the Extensions3D
object or its subclasses.

Tested:
 - Chromium on Linux with accelerated 2D canvas and HTML5 video
 - Chromium on Mac OS X with WebGL and CSS 3D content
 - Safari on Mac OS X with WebGL and CSS 3D content

No new tests. Covered by existing tests.