2006337 - (css-anchor-position-interop-2025) [css-anchor-position] Investigate remaining interop 2025 failures

Status: NEW

(Core :: Layout: Positioned, defect, P3)

Description

[Emilio Cobos Álvarez [:emilio]]

Link: https://wpt.fyi/results/css/css-anchor-position?label=master&label=experimental&aligned&view=interop&q=label%3Ainterop-2025-anchor-positioning%20and%20firefox%3Afail

My understanding is:

Multicol

• anchor-getComputedStyle-002.html
• anchor-position-multicol*

position-try-order (bug 1989059)

• All the ones in parsing/
• property-interpolations.html failures
• position-try-order-basic.html
• position-try-order-position-area.html

@position-try invalidation (bug 1910616)

• at-position-try-cssom.html failure
• at-position-try-invalidation-shadow-dom.html (maybe, may depend on scoped names too)
• at-position-try-invalidation.html

position-try-fallbacks should use the IMCB (bug 2004495 + bug 1924792)

• A bunch in position-try-*
• position-visibility-remove-no-overflow.html
• Probably others

Last successful position-fallback selection (bug 2006373)

(Note these might also depend on ^)

• last-successful-*

Scrolling (bug 2002789 + another one tbd)

• anchor-scroll-chained-{002,003}.html (bug 2002789)
• anchor-scroll-position-try-*

Should be removed (no bug)

• anchor-inherited.html: https://github.com/web-platform-tests/interop/issues/1235

Depends on interleaving, but can be fixed (bug 2006368)

• try-tactic-sizing.html (we should change it to test rendering)

calc() simplification (bug 2006366)

• anchor-parse-valid.html (probably can be tweaked since problem seems unrelated to anchor() itself)

Tree scoping of names (bug 1992387)

• anchor-scope-shadow-flat-tree.html
• position-try-tree-scoped.html
• anchor-scope-shadow-*
• anchor-name-cross-shadow.html

Grid interaction (bug 2004596)

• anchor-position-grid-001.html
• position-area-in-grid.html
• position-try-grid-001.html

Others / need more digging

• auto-margins-position-area.html
• position-area-basic.html
• position-try-backdrop.html (probably depends on ::backdrop being a real element...)
• Any other that I haven't mentioned

Comment 1

[David Shin[:dshin]]

Quick comments:

• anchor-parse-valid.html: We simplify anchor(min(50%, 100%)) to anchor(calc(50%)) which the test does not expect.
• anchor-position-grid-001.html: It has multicol mixed in it :(

Comment 2

[Emilio Cobos Álvarez [:emilio]]

Can the percentage basis of anchor() be negative? If not, our behavior seems more correct.

Comment 3

[David Shin[:dshin]]

(In reply to Emilio Cobos Álvarez [:emilio] from comment #2)

Can the percentage basis of anchor() be negative? If not, our behavior seems more correct.

The basis is absolute containing block's size in that axis, which should be non-negative.
FWIW: calc() spec on simplification does say:

If a percentage is left at this point, it will usually block simplification of the node

Comment 4

[David Shin[:dshin]]

position-try-fallbacks should use the IMCB

Bug 2004495

position-visibility-remove-no-overflow.html

Also depends on bug 1924792

Comment 5

[David Shin[:dshin]]

anchor-scroll-chained-{002,003}.html

Bug 2002789

Comment 6

[David Shin[:dshin]]

position-try-001.html (And maybe others as well) may depend on bug 2006373.
That test fails for us because we use the base style after trying out all options, but also we end up using the last fallback option's size.

Comment 7

[Daniel Holbert [:dholbert]]

For convenience, here's wpt.fyi showing an up-to-date list of interop-2025-relevant Firefox Nightly test-failures here:
https://wpt.fyi/results/css/css-anchor-position?label=master&label=experimental&product=chrome&product=firefox&product=safari&product=edge&aligned&view=interop&q=label%3Ainterop-2025-anchor-positioning%20firefox%3A%21pass%20firefox%3A%21ok

[EDIT: ah, sorry, I missed that we had a query in comment 0 as well. Note that the two queries are slightly different [I copypasted the one from an internal tracking spreadsheet] but the test list is the same, so no news here.]