280958 2024-10-06 22:58:15 -0700 [site-isolation] RenderLayerBacking::updateAfterWidgetResize needs to notify cross-process subframe RenderLayerCompositors of bounds changing 2024-10-13 14:00:19 -0700 1 1 1 Unclassified WebKit Layout and Rendering WebKit Nightly Build Unspecified Unspecified RESOLVED WONTFIX InRadar P2 Normal --- 281269 1 mattwoodrow mattwoodrow bfulgham commit-queue simon.fraser webkit-bug-importer zalan oldest_to_newest 2065778 0 mattwoodrow 2024-10-06 22:58:15 -0700 The code currently only works with in-process subframes. 2065779 1 webkit-bug-importer 2024-10-06 22:58:25 -0700 <rdar://problem/137406024> 2065780 2 mattwoodrow 2024-10-06 23:01:36 -0700 Pull request: https://github.com/WebKit/WebKit/pull/34756 2066678 3 ews-feeder 2024-10-09 15:37:18 -0700 Committed 284932@main (3f1856724b13): <https://commits.webkit.org/284932@main> Reviewed commits have been landed. Closing PR #34756 and removing active labels. 2067088 4 commit-queue 2024-10-10 18:50:36 -0700 Re-opened since this is blocked by bug 281269 2067611 5 mattwoodrow 2024-10-13 14:00:19 -0700 I reverted this because it causes visual rendering issues in http/tests/site-isolation/selection-focus.html, despite making the graphics layer dumps match. It looks like in the site-isolation case, the RenderLayer for the <iframe> creates a 'contents' GraphicsLayer, and the relevant offset is set on that (and that layer isn't included in the dump). For in-process iframes, there's no contents layer, just the child RenderView/RenderLayerCompositor, and we manually forward the offset down to it. I don't think there's anything to fix here, unless we really need the layer trees to be identical (which is nice for testing, but probably doesn't matter).