RESOLVED FIXED 107224
Coordinated Graphics: LayerTreeRenderer manages the surface of UpdateAtlas. 
https://bugs.webkit.org/show_bug.cgi?id=107224
Summary Coordinated Graphics: LayerTreeRenderer manages the surface of UpdateAtlas.
Dongseong Hwang
Reported 2013-01-17 21:53:27 PST
Currently, CoordinatedLayerTreeHostProxy manages the surface of UpdateAtlas, but all other resources are managed by LayerTreeRenderer. This patch matches the surface of UpdateAtlas to other resources.
Attachments
Patch (8.70 KB, patch)
2013-01-17 21:54 PST, Dongseong Hwang 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Dongseong Hwang
Comment 1 2013-01-17 21:54:51 PST
Created attachment 183370 [details] Patch
Dongseong Hwang
Comment 2 2013-01-17 21:57:12 PST
CoordinatedLayerTreeHostProxy::m_surface is obstacle that TextureMapperScene handles all resources, so I move m_surface from CoordinatedLayerTreeHostProxy to LayerTreeRenderer.
Dongseong Hwang
Comment 3 2013-01-24 21:31:59 PST
noam: ping
Dongseong Hwang
Comment 4 2013-01-27 22:05:20 PST
Could noam review please?
Noam Rosenthal
Comment 5 2013-01-27 23:02:44 PST
Comment on attachment 183370 [details] Patch ... why is this needed? I think it makes sense for the web process to manage the lifestyle of update atlases.
Dongseong Hwang
Comment 6 2013-01-27 23:14:27 PST
(In reply to comment #5) > (From update of attachment 183370 [details]) > ... why is this needed? > I think it makes sense for the web process to manage the lifestyle of update atlases. web process still manage the lifecycle. It just moves surface map from CoordinatedLayerTreeHostProxy to LayerTreeRenderer, because other resources (e.g. ImageBacking, Canvas Surface, CoordinatedTiledBackingStore, etc.) belong to LayerTreeRenderer. On the other hands, I don't want CoordinatedLayerTreeHostProxy to have this kind of code because we will remain only enqueueCoordinatedOperation and commitCoordinatedCoperations in CoordinatedLayerTreeHostProxy.
Noam Rosenthal
Comment 7 2013-01-27 23:23:01 PST
Comment on attachment 183370 [details] Patch LGTM
Noam Rosenthal
Comment 8 2013-01-27 23:23:22 PST
(In reply to comment #6) > (In reply to comment #5) > > (From update of attachment 183370 [details] [details]) > > ... why is this needed? > > I think it makes sense for the web process to manage the lifestyle of update atlases. > > web process still manage the lifecycle. > It just moves surface map from CoordinatedLayerTreeHostProxy to LayerTreeRenderer, because other resources (e.g. ImageBacking, Canvas Surface, CoordinatedTiledBackingStore, etc.) belong to LayerTreeRenderer. > > On the other hands, I don't want CoordinatedLayerTreeHostProxy to have this kind of code because we will remain only enqueueCoordinatedOperation and commitCoordinatedCoperations in CoordinatedLayerTreeHostProxy. Right, I was misunderstanding the patch (haven't had coffee yet...)
Dongseong Hwang
Comment 9 2013-01-27 23:28:17 PST
(In reply to comment #8) > Right, I was misunderstanding the patch (haven't had coffee yet...) hehe :)
Dongseong Hwang
Comment 10 2013-01-29 23:45:54 PST
(In reply to comment #7) > (From update of attachment 183370 [details]) > LGTM benjaminp, could you take a look?
Benjamin Poulain
Comment 11 2013-01-30 14:46:55 PST
Comment on attachment 183370 [details] Patch I am okay with this and Noam reviewed.
WebKit Review Bot
Comment 12 2013-01-30 15:01:52 PST
Comment on attachment 183370 [details] Patch Clearing flags on attachment: 183370 Committed r141325: <http://trac.webkit.org/changeset/141325>
WebKit Review Bot
Comment 13 2013-01-30 15:01:57 PST
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.