1919358 - [css-view-transitions] Implement invalidation for view transition pseudo-elements

Status: RESOLVED WORKSFORME

(Core :: CSS Parsing and Computation, task, P3)

Comment 1

[Emilio Cobos Álvarez [:emilio]]

Changing a root class should change the styles of pseudo-elements as needed (like here)

Comment 2

[Emilio Cobos Álvarez [:emilio]]

So I think given how boris implemented the parsing, if we implement the selector-matching correctly, and the traversal in a way such that the view transition pseudo tree is under the root, then I think this "just" works, but otherwise it probably needs some amount of work in the invalidation code.

Comment 3

[Boris Chiou [:boris]]

It seems the test (https://searchfox.org/mozilla-central/source/testing/web-platform/tests/css/css-view-transitions/pseudo-rendering-invalidation.html) is still not passed. I'm trying to debug it.

Comment 4

[Boris Chiou [:boris]]

(In reply to Boris Chiou [:boris] from comment #3)

It seems the test (https://searchfox.org/mozilla-central/source/testing/web-platform/tests/css/css-view-transitions/pseudo-rendering-invalidation.html) is still not passed. I'm trying to debug it.

Well. Looks like I cannot reproduce it locally today. It always passes. Let's check in try server again:
https://treeherder.mozilla.org/jobs?repo=try&revision=a8446261008e593d1300614a9c1f3556fd712888

Comment 5

[Boris Chiou [:boris]]

Attachment: reftest difference

Looks like there is one difference in each color channel component, in "Group" and "Image-Pair". It's hard to see the difference with my eyes actually.

I guess this intermittent is due to some other rendering issues, not the style invalidation. If the invalidation doesn't work, it should have a significant difference (because we change from display:none in the test).

Comment 6

[Boris Chiou [:boris]]

So I think we do invalidation already, but the failures is due to Bug 1950763.

Comment 7

[Boris Chiou [:boris]]

So there is nothing we can do in this bug. We could update the test expectation of pseudo-rendering-invalidation.html in Bug 1839326 once we fix the rendering issue or find a tentative way to update the test.

Comment 8

[Boris Chiou [:boris]]

So looks like we still have to implement this (per bug 1976539). We accidentally pass other tests because we always restyle all the pseudo-elements when activating the view transition (because we just insert these pseudo-elements). Anyway, I am adding invalidation code in Bug 1976539 (so we don't need to reopen this).