1928223 - Firefox differs from Chrome on float stacking order, in relatively-positioned inline with ib-split sibling

Status: NEW

(Core :: Layout: Floats, defect)

Description

[Daniel Holbert [:dholbert]]

Attachment: testcase 1

This is a reduced testcase from bug 1926971, though I'm not yet sure whether this is the main issue in that bug.

STR:

1. Load attached testcase.

ACTUAL RESULTS:
Top half of blue rect is covered up by the lightgray rect.

EXPECTED RESULTS:
Top half of blue rect should (maybe?) be visible, stacking on top.

(I'm not entirely sure who's correct here; painting order has a lot of subtlety, particularly here where we've got a mix of floats, positioning, and a block-in-inline-split.)

Firefox & WebKit (epiphany) give ACTUAL RESULTS.
Chrome gives EXPECTED RESULTS.

Our behavior goes back as far as I can test -- Nightly 2009-12-31 3.7a1pre matches current Nightly.

Comment 1

[Daniel Holbert [:dholbert]]

Chrome 100 and earlier match Firefox/WebKit here (top half of blue div covered up).
Chrome 101 and newer match current Chrome (top half of blue div is visible/foreground).

Comment 2

[Daniel Holbert [:dholbert]]

More specifically: Chrome 100 behaves like WebKit on the testcase here and on both testcases in bug 1926971 -- they match Firefox on Jeff's "reduced test case" but not on the "standalone test case" on that bug.

So: this bug here is really a reduction of the "reduced testcase" which demonstrates a difference vs. Chrome 101-and-beyond, which is independent from our failure in bug 1926971.

Comment 3

[Daniel Holbert [:dholbert]]

(This seems similar to bug 1899829.)