1650885 - Exported grid container baseline should use align-self:baseline item baseline, not first item baseline

Status: NEW

(Core :: Layout: Grid, defect)

Description

[fantasai]

When some grid items in a grid row are baseline-aligned, the baseline of those items (rather than the first item in grid order) should be used for baseline alignment of the grid container.

Spec: https://drafts.csswg.org/css-grid-1/#grid-baselines
Test: wpt /css/css-grid/alignment/grid-baseline-004.html

Comment 2

[Daniel Holbert [:dholbert]]

This still seems to be a valid bug, and we still fail the associated wpt test because of it:
https://wpt.fyi/results/css/css-grid/alignment/grid-baseline-004.html
http://wpt.live/css/css-grid/alignment/grid-baseline-004.html

There are a handful of failures, but the first/easiest to see (if you scroll down to the bottom of the test) is in the second .container black box. It has three inline-grids (baseline-aligned in a block), and we're using the first grid item in each of those grids to determine the grid's baseline.

We should instead be using the baseline-aligned item in each one to establish its baseline, which is the blue item (not necessarily the first).

Comment 3

[Daniel Holbert [:dholbert]]

(note: bug 1609403 is also associated with some interop-2023-grid test failures related to baseline alignment, but that one's focused on grid items getting the correct baseline-aligned placement, whereas this one here is focused on the grid container determining its own baseline from its items)

Comment 4

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #2)

This still seems to be a valid bug, and we still fail the associated wpt test because of it:
https://wpt.fyi/results/css/css-grid/alignment/grid-baseline-004.html
http://wpt.live/css/css-grid/alignment/grid-baseline-004.html

There are a handful of failures

Update: there's only one failure in this test now -- we fail subtest #17 of 18, which is about the offsetTop of the middle grid in the bottom-most section.

The failure still seems to be related to us not quite matching the test's expectation for the exported baseline.