1833486 - Firefox fails wpt test css-grid/subgrid/subgrid-stretch.html

Status: RESOLVED FIXED

(Core :: Layout: Grid, defect)

Description

[Daniel Holbert [:dholbert]]

wpt.fyi: https://wpt.fyi/results/css/css-grid/subgrid/subgrid-stretch.html

wpt.live:
test: http://wpt.live/css/css-grid/subgrid/subgrid-stretch.html
reference: http://wpt.live/css/css-grid/subgrid/subgrid-stretch-ref.html

It looks like the inline-grid elements are all 100x100 squares, and the test is expecting the blue subgrid inside to stretch to fill that area, per this spec quote that the test links to:

The subgrid is always stretched in its subgridded dimension(s): the align-self/justify-self properties on it are ignored, as are any specified width/height constraints.

https://drafts.csswg.org/css-grid-2/#subgrid-box-alignment

Comment 1

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

Attachment: Bug 1833486 - Implement stretch for subgrids in the subgridded axis. r?dholbert

This implements https://drafts.csswg.org/css-grid/#subgrid-box-alignment

In subgrid-stretch.html wpt, those inline-grid containers are wrapped into
multiple lines. Firefox might have a baseline alignment bug such that there is
no gap between line 3 and line 4. Since the wpt is meant to test subgrid
stretch, not grid container's baseline, I add vertical-align:top to make
Firefox pass the test and margin-bottom for better visualization.

Depends on D196705

Comment 3

[Phabricator Automation]

Comment on attachment 9369160 [details]
Bug 1833486 - Implement stretch for subgrids in the subgridded axis. r?dholbert

Revision D196706 was moved to bug 1841296. Setting attachment 9369160 [details] to obsolete.

Comment 4

[Daniel Holbert [:dholbert]]

TYLin's patch which just moved to bug 1841296 gets us nearly to passing this test.

There's just one minor cosmetic difference here that's still causing a test-failure: the test's final inline-grid on each of the two lines happens to linewrap to a new line, and that final linewrapped inline-grid is vertically a bit higher in the testcase vs. the reference case, due to the fact that we apparently export a central baseline for grids with an orthogonal-writing-mode grid item. (That seems incorrect.)

Comment 5

[Daniel Holbert [:dholbert]]

Attachment: reduced testcase for baseline issue

This is a reduced testcase for the baseline-alignment issue. The root cause might be tracked in another interop-2023 issue already (we have a handful of remaining baseline-related grid/subgrid test failures).

Comment 6

[Daniel Holbert [:dholbert]]

This test will starts passing once my patch (pending review) lands over in bug 1870949. (It just fixes the last minor issue; TYLin's fix in bug 1841296 fixed the main issue here).

Comment 7

[Daniel Holbert [:dholbert]]

That patch landed and this test is indeed showing up as passing now here:
https://wpt.fyi/results/css/css-grid/subgrid/subgrid-stretch.html?label=experimental&label=master

(This isn't yet reflected in interop-2023 dashboard, just because we don't yet have an "aligned run" [with results for all browsers] that includes the fix.)

Resolving as FIXED by the patches that landed in bug 1870949 and bug 1841296. (This isn't exactly a dupe of either one of them; it depended on both.)