Summary: | Layers should be destroyed by RenderLayerModelObject | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Antti Koivisto <koivisto> | ||||
Component: | Layout and Rendering | Assignee: | Nobody <webkit-unassigned> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | bfulgham, commit-queue, simon.fraser, webkit-bug-importer, zalan | ||||
Priority: | P2 | Keywords: | InRadar | ||||
Version: | WebKit Nightly Build | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Antti Koivisto
2017-10-10 11:51:41 PDT
Created attachment 323328 [details]
patch
Comment on attachment 323328 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=323328&action=review > Source/WebCore/rendering/RenderObject.cpp:-1444 > - // FIXME: Would like to do this in RenderBoxModelObject, but the timing is so complicated that this can't easily > - // be moved into RenderLayerModelObject::willBeDestroyed(). > - // FIXME: Is this still true? > - if (hasLayer()) { > - setHasLayer(false); > - downcast<RenderLayerModelObject>(*this).destroyLayer(); > - } > - Should RenderObject have ASSERT(!hasLayer()) somewhere late in its destruction to ensure that we didn't re-assign a layer after RenderLayerModelObject::destroyLayer() ran? There isn't really much between willBeDestroyed() and the destructor. Comment on attachment 323328 [details] patch Clearing flags on attachment: 323328 Committed r223139: <http://trac.webkit.org/changeset/223139> All reviewed patches have been landed. Closing bug. |