1848446 - Firefox fails WPT css/css-grid/alignment/grid-align-baseline-table-002.html

Status: NEW

(Core :: Layout: Grid, defect)

Description

[Daniel Holbert [:dholbert]]

We fail subtests in these interop-2023-grid WPT tests:
https://wpt.fyi/results/css/css-grid/alignment/grid-align-baseline-table-001.html
https://wpt.fyi/results/css/css-grid/alignment/grid-align-baseline-table-002.html
https://wpt.fyi/results/css/css-grid/alignment/grid-align-baseline-table-003.html

wpt.live links:
http://wpt.live/css/css-grid/alignment/grid-align-baseline-table-001.html
http://wpt.live/css/css-grid/alignment/grid-align-baseline-table-002.html
http://wpt.live/css/css-grid/alignment/grid-align-baseline-table-003.html

As in bug 1848424, it looks like all of the failures are in cases where we're doing last baseline alignment, and we're failing to extract a last baseline from the grid item (which happens to be a table).

Some additional references:

• the results table in these tests are a bit broken-looking right now; that's not relevant to our pass/fail status but it'll be fixed in bug 1848445.
• these tests also use multiple table-captions which we don't support (bug 144517). I haven't checked whether that'll mean we'll still fail these tests or not when we fix the last baseline alignment issue; but it's possible we'll still fail a subtest or two due to that.

Comment 1

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #0)

As in bug 1848424, it looks like all of the failures are in cases where we're doing last baseline alignment, and we're failing to extract a last baseline from the grid item (which happens to be a table).

Note: it's possible that bug 1848424's fix will fix this bug too (or vice versa), if the issue is in some common grid code that's failing to correctly request a last baseline or something.

Comment 2

[Tiaan Louw [:tlouw]]

-001 and -003 were fixed by bug 1876612

Comment 3

[Daniel Holbert [:dholbert]]

https://wpt.live/css/css-grid/alignment/grid-align-baseline-table-002.html is still failing two subtests, but I confirmed that it's just a version of bug 144517 as alluded to at the end of comment 0.

(The test is expecting offsetLeft values that are 40px larger than what we're getting ("offsetLeft expected 115 but got 75", "offsetLeft expected 65 but got 25"). If I add view the testcase in Chrome and add display:none to the bottom caption (to make Chrome ignore those captions in the same way that they're ignored in Firefox), then they produce the same offsetLeft values that we do for those elements (75 and 25). So the 40px difference just comes from the width of that "Bottom" caption which we don't render.)