So, test_pointerPreserves3DPerspective.html fails because of [MOZ_ASSERT(aChild->GetApzc() != parent) in HitTestingTreeNode::SetLastChild](https://searchfox.org/mozilla-central/rev/f91bd38732d4a330eba4e780812274b98eb81274/gfx/layers/apz/src/HitTestingTreeNode.cpp#88), this is pretty much same as bug 1458145. As per a comment by Botond in bug 1458145 comment 4, it means the layer tree is malformed. Though I don't quite understand how ASR stuff works, my speculation is that we need to preserve the current ASR if we couldn't find the appropriate ASR for perspective item, list [this change](https://hg.mozilla.org/try/rev/c48b7637b2225c9d07e450aaddc5ecadaef03e7a). Markus, does this change reasonable? The change also fixes the first test case in bug 1458145 (the second one doesn't cause the assertion now), so I am hoping I am on the right track. That's said, the change makes [perspective-scrolling-3.html fail on non-E10S Android](https://treeherder.mozilla.org/#/jobs?repo=try&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel%2Crunning%2Cpending%2Crunnable&revision=65d2d073f06d0249e2da8f7a55c542955bd214ff&selectedJob=253872664) (the test passes on E10S Android), so there needs some more work. But I guess it's due to another pre-existing issue on reftest-async-scroll or relevant stuff. CCing Botond, he might have some ideas about the perspective-scrolling-3.html failure.
Bug 1523500 Comment 8 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
So, test_pointerPreserves3DPerspective.html fails because of [MOZ_ASSERT(aChild->GetApzc() != parent) in HitTestingTreeNode::SetLastChild](https://searchfox.org/mozilla-central/rev/f91bd38732d4a330eba4e780812274b98eb81274/gfx/layers/apz/src/HitTestingTreeNode.cpp#88), this is pretty much same as bug 1458145. As per a comment by Botond in bug 1458145 comment 4, it means the layer tree is malformed. Though I don't quite understand how ASR stuff works, my speculation is that we need to preserve the current ASR if we couldn't find the appropriate ASR for perspective item, like [this change](https://hg.mozilla.org/try/rev/c48b7637b2225c9d07e450aaddc5ecadaef03e7a). Markus, does this change reasonable? The change also fixes the first test case in bug 1458145 (the second one doesn't cause the assertion now), so I am hoping I am on the right track. That's said, the change makes [perspective-scrolling-3.html fail on non-E10S Android](https://treeherder.mozilla.org/#/jobs?repo=try&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel%2Crunning%2Cpending%2Crunnable&revision=65d2d073f06d0249e2da8f7a55c542955bd214ff&selectedJob=253872664) (the test passes on E10S Android), so there needs some more work. But I guess it's due to another pre-existing issue on reftest-async-scroll or relevant stuff. CCing Botond, he might have some ideas about the perspective-scrolling-3.html failure.
So, test_pointerPreserves3DPerspective.html fails because of [MOZ_ASSERT(aChild->GetApzc() != parent) in HitTestingTreeNode::SetLastChild](https://searchfox.org/mozilla-central/rev/f91bd38732d4a330eba4e780812274b98eb81274/gfx/layers/apz/src/HitTestingTreeNode.cpp#88), this is pretty much same as bug 1458145. As per a comment by Botond in bug 1458145 comment 4, it means the layer tree is malformed. Though I don't quite understand how ASR stuff works, my speculation is that we need to preserve the current ASR if we couldn't find the appropriate ASR for perspective item _in the same document_, like [this change](https://hg.mozilla.org/try/rev/c48b7637b2225c9d07e450aaddc5ecadaef03e7a). Markus, does this change reasonable? The change also fixes the first test case in bug 1458145 (the second one doesn't cause the assertion now), so I am hoping I am on the right track. That's said, the change makes [perspective-scrolling-3.html fail on non-E10S Android](https://treeherder.mozilla.org/#/jobs?repo=try&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel%2Crunning%2Cpending%2Crunnable&revision=65d2d073f06d0249e2da8f7a55c542955bd214ff&selectedJob=253872664) (the test passes on E10S Android), so there needs some more work. But I guess it's due to another pre-existing issue on reftest-async-scroll or relevant stuff. CCing Botond, he might have some ideas about the perspective-scrolling-3.html failure.