1095937 - improve inline-sizing of orthogonal flows

Status: RESOLVED FIXED

(Core :: Layout: Block and Inline, defect)

Description

[Jonathan Kew [:jfkthame]]

Attachment: testcase with an orthogonal flow within a small <div>

When we reflow an orthogonal flow, we currently apply an mOrthogonalLimit from a parent reflow state that is based on the top-level viewport size. This leads to poor results when the orthogonal flow's parent is itself a smaller block, not the page as a whole. See attached testcase.

Comment 1

[Jonathan Kew [:jfkthame]]

Attachment: Apply better inline-size limits to orthogonal flows.

Comment 2

[Jonathan Kew [:jfkthame]]

Comment on attachment 8519464 [details] [diff] [review]
Apply better inline-size limits to orthogonal flows.

Cancelling r? for now, as this appears to have a bad effect on other examples I've been playing with. Needs further investigation.

Comment 3

[fantasai]

Attachment: testcase

Hi jfkthame,
I'm not sure what you were trying to do here, but css-writing-modes has some fairly specific rules about orthogonal flow sizing in https://www.w3.org/TR/css-writing-modes-3/#orthogonal-flows

I've attached a testcase which shows some interesting ways in which the current mOrthogonalLimit system fails. In particular the code you're trying to modify in Initi() doesn't work if the frame that has an mOrthogonalLimit set doesn't happen to be orthogonal to an orthogonal flow (because we've nested orthogonal writing modes). You can see this case in the 'auto' example of the testcase, which ends up sizing as max-content instead of as the ICB height due to the nesting.

Comment 4

[fantasai]

Attachment: Bug 1095937 - Improve inline-sizing of orthogonal flows: deal with nesting and match to css-writing-modes-3 spec.

Comment 5

[fantasai]

This patch brings us in line with css-writing-modes-3 wrt orthogonal limits.

(It doesn't handle resizing of the ICB correctly, but that's an existing problem. Fixing that requires invalidating layout of the orthogonal flow at appropriate times, which is a bit tricky since the frame relationship is quite indirect, which is probably why we don't do it right now. I can try to figure that out, but I'd prefer to land this first and handle that separately.)

Comment 6

[BugBot [:suhaib / :marco]]

The bug assignee is inactive on Bugzilla, so the assignee is being reset.

Comment 7

[Ting-Yu Lin [:TYLin] (PST, UTC-7)]

Jonathan, this bug came to mind. Do you feel fantasai's patch might help bug 1310551?

Comment 8

[Jonathan Kew [:jfkthame]]

I think the approach in D45948 isn't quite right, as it "accumulates" constraints from successive scroll frames, but the spec actually says to consider only the closest ancestor scroll frame. That led to the WPT failure fantasai noted in https://phabricator.services.mozilla.com/D45948#1524904. The patch also didn't account for a fixed size of the immediate containing block, and as a result it didn't fix examples like bug 1949064.

I'm landing an alternative patch in bug 1949064 that should fix these issues. I notice that fantasai's patch also included some new WPT reftests, so we should extract those and add them to the suite, for increased test coverage.

Comment 9

[Jonathan Kew [:jfkthame]]

Attachment: Bug 1095937 - Add new WPT reftests for orthogonal-flow available-size constraints. r=#layout

These are the new tests that were included in fantasai's patch D45948.
With the patch that landed in bug 1949064 they should all now pass.

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/bb95e001c526
Add new WPT reftests for orthogonal-flow available-size constraints. r=layout-reviewers,emilio

Comment 11

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/51315 for changes under testing/web-platform/tests

Comment 12

[Cristian Tuns]

https://hg.mozilla.org/mozilla-central/rev/bb95e001c526

Comment 13

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Upstream PR merged by moz-wptsync-bot