[webkit-reviews] review requested: [Bug 79927] [chromium] Add clipping to scissor rect to CCOcclusionTracker : [Attachment 129912] Patch

[bugzilla-daemon at webkit.org]

Dana Jansens <danakj at chromium.org> has asked  for review:
Bug 79927: [chromium] Add clipping to scissor rect to CCOcclusionTracker
https://bugs.webkit.org/show_bug.cgi?id=79927

Attachment 129912: Patch
https://bugs.webkit.org/attachment.cgi?id=129912&action=review

------- Additional Comments from Dana Jansens <danakj at chromium.org>
Ok this is converging I hope.

1) Uncomfortable with doing a different scissor calculation, I changed the
interface so that CCLTHI can give the
same damage rects to the occlusionTracker that LRC will be using for drawing. I
did this with a damageClient interface,
so CCOcclusionTracker can just query the client which will give back the
damage. This lets us avoid knowing about things
 like CCRenderPass etc. And it can be null to not use damage.

1b) Also the client pattern here has a perk that we can still move the
for-each-layer logic from CCLTH(I) into the
occlusionTracker in the future if we should choose to.

2) I realized that your uncertainty about computing occlusion of surfaces was
well founded. While the
m_owningLayer->screenSpaceTransform seemed to do well enough for the surface
contents, it would not include
the surface replica, and if that's not occluded, we need to consider the
surface unoccluded. So I just took
out the surface methods, and we'll just worry about layers like we do currently
in CCQuadCuller.

3) Added some more unit tests to test the interaction with the damage rect,
instead of the overall scissor rects.