Reconsideration: allow `abs-sign` to advance to Baseline low

[ddbeck]

In #3189, I put an editorial hold on abs-sign, preventing it from advancing to Baseline low. We should reconsider this decision.

Background

When reviewing #3189, I saw that this feature was advancing. According to the latest BCD, this feature shipped most recently in Chrome 138. I checked for open bugs against BCD and the major browser vendor issue trackers that might touch on this feature. I found https://bugzilla.mozilla.org/show_bug.cgi?id=1845298, which reports an issue with mixed relative units in sign().

After digging into that, I found some potentially related bugs (to do with mixed units) that were not or are not reported in BCD:

• https://bugzilla.mozilla.org/show_bug.cgi?id=1928480
• https://bugzilla.mozilla.org/show_bug.cgi?id=1827404
• https://bugs.webkit.org/show_bug.cgi?id=255280
• https://issues.chromium.org/issues/40768696

Options

We could handle this in a few ways:

• Advance: Undo my editorial override and let the BCD change go through. When this comes up again at the low→high boundary, we'll have a chance to think about this again, if the bug remains. We should choose this option if we think the behavior is unimportant to developers.
• Submit changes upstream: Add data to BCD reflecting the bug in Firefox and the fixed bugs in Safari and Chrome. When those changes land in BCD, undo the editorial override here and follow BCD again. We should choose this option if we think the behavior is important to developers.
• Keep the editorial override in place: silently ignore BCD. Note that this does not override per-key statuses. We should choose this option if we cannot come to an agreement about the other options.

[jgraham]

CC @emilio : do we think this Firefox bug is an edge case or something developers are likely to run into? It seems like so far there aren't any webcompat issues blocked on the gecko bug, but I don't have a good sense of whether that's just because the feature isn't widely used yet.

[emilio]

The mixed units things are basically a tangential feature, IMO (allowing length units on non-length properties, and do unit math in calc()).

[ddbeck]

Thanks for the input, @emilio.

We discussed this at the WebDX CG meeting on July 31 and nobody voiced opposition to allowing the feature to advance. I'm also satisfied that this hold is not needed, so I've sent #3205 to remove the editorial hold on this feature and progress it to Baseline low.