1626212 - New wpt failures from PR 22445: css/css-flexbox/content-height-with-scrollbars.html

Status: RESOLVED INVALID

(Core :: Layout: Flexbox, defect)

Description

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

The following tests have untriaged failures in the CI runs for wpt PR 22445:

New Tests That Don't Pass

/css/css-flexbox/content-height-with-scrollbars.html: FAIL [Gecko-android-em-7.0-x86_64-debug-geckoview, Gecko-android-em-7.0-x86_64-opt-geckoview], PASS [Gecko-linux1804-64-asan-opt, Gecko-linux1804-64-debug, Gecko-linux1804-64-opt, Gecko-linux1804-64-qr-debug, Gecko-linux1804-64-qr-opt, Gecko-windows10-64-debug, Gecko-windows10-64-opt, Gecko-windows10-64-qr-debug, Gecko-windows10-64-qr-opt, Gecko-windows7-32-debug, Gecko-windows7-32-opt, GitHub] (Chrome: PASS, Safari: PASS)

These updates will be on mozilla-central once bug 1624948 lands.

Note: this bug is for tracking fixing the issues and is not
owned by the wpt sync bot.

This bug is linked to the relevant tests by an annotation in
https://github.com/web-platform-tests/wpt-metadata. These annotations
can be edited using the wpt interop dashboard
https://jgraham.github.io/wptdash/

If this bug is split into multiple bugs, please also update the
annotations, otherwise we are unable to track which wpt issues are
already triaged. Resolving as duplicate or closing this issue should
be cause the bot to automatically update or remove the annotation.

Comment 2

[Daniel Holbert [:dholbert]]

Live link to test & ref:
http://wpt.live/css/css-flexbox/content-height-with-scrollbars.html
http://wpt.live/css/css-flexbox/reference/content-height-with-scrollbars-ref.html

The failure here seems to be Android-specific, per comment 0. It's because the test is incorrectly assuming that scrollbars take up some nonzero amount of content-height (and they don't, on Android where they're "overlay"-style, in Firefox at least).

The problem is here, in the 3rd flexbox in the test:

<div class="flexbox column wrap">
  <div style="height: 50px; width: 50%;"></div>
  <div style="height: 50px; width: 50%; background-color: lightblue"></div>
</div>

The outer flexbox here has height:100px; overflow: scroll

The test is asserting that the two 50px-tall children do not fit vertically into that 100px of height, & it's expecting they end up side by side (wrapping to be alongside each other). This makes sense if there's a scrollbar at the bottom of the flexbox that steals some of its 100px of height -- but if scrollbars are overlay-style & occupy no space, then the two children stack vertically just fine.

So: this part of the test has bogus assumptions, on platforms with overlay scrollbars, and the test needs to be rewritten to avoid/remove this invalid assumption.

Comment 3

[Daniel Holbert [:dholbert]]

The 4th flexbox has the same issue, too. (In the test failure, both the 3rd and the 4th render as green-above-blue, instead of green-next-to-blue).

The test fails in the same way on Firefox-for-Android and Chrome-for-Android, too, so this isn't a Firefox-specific failure.

Comment 4

[Daniel Holbert [:dholbert]]

I posted about this on the pull request, in case the folks involved there want to backout or anything: https://github.com/web-platform-tests/wpt/pull/22445#issuecomment-606810015

In the meantime, I think we're fine to annotate the test as failing-on-Android (to reflect reality, per comment 0).

Comment 5

[Daniel Holbert [:dholbert]]

(reopening, because I think this was duped by accident. Our automation did seem to file a bunch of redundant bugs for the same test, but this one and current-dupe-target bug 1625762 are about different tests.)

Comment 6

[Christian :Biesinger (don't email me, ping me on IRC)]

I haven't analyzed this test, but that does mean that this likely also fails in Chrome on Mac when run manually (our testrunner disables overlay scrollbars)

Comment 10

[Antonio Gomes (tonikitoo)]

PS: Not sure if "Antonio Gomez" was added to cc/ by accident. It'd not be the first time of a Gomes x Gomez confusion. :)

Comment 11

[Daniel Holbert [:dholbert]]

Yes that's definitely my fault. Apologies to both of you. :) I'll un-CC him.

Comment 12

[Daniel Holbert [:dholbert]]

And I guess I'll close this as INVALID since this was a bug in the test rather than in Firefox, per comment 2 - comment 4.