Bug 247826 – Serialization differences with "columns" with one or more `auto` value

• Component: CSS
• Status: RESOLVED FIXED

[Ahmad Saleem]

Hi Team,

While going through Bug 103597 and trying to see "column-count: auto" support, I came across following Blink Commit:

Blink Commit - https://src.chromium.org/viewvc/blink?view=revision&revision=155047

I took the test case from above commit and changed it into "JSFiddle" below:

Link - https://jsfiddle.net/2jhw4k5x/show

It seems that we fail following tests compared to Chrome Canary 109:

FAIL element.style.WebkitColumns should be auto auto. Was auto.
FAIL element.style.WebkitColumns should be auto auto. Was auto.
FAIL element.style.WebkitColumns should be auto auto. Was auto.
FAIL element.style.WebkitColumns should be auto 7. Was 7.
FAIL element.style.WebkitColumns should be 7em auto. Was 7em.

Rather than opening old bug, I just thought to create a separate to track this issue. Thanks!

[Tim Nguyen (:ntim)]

This was already fixed in https://commits.webkit.org/182834@main

The remaining cases you reported are about serialization differences, what we do is correct given the guideline is always to use the shortest form.

auto auto -> auto
auto 7 -> 7
7em auto -> 7em.

Seems reasonable IMO as I don't see a case where this could cause ambiguity, but we can always adjust.

Firefox seems to be like Chrome and always serializes both for the shorthand: https://jsfiddle.net/3nd4w5qh/show

Oriol, Emilio, wdyt?

[Oriol Brufau]

Yes I would say that omitting optional values seems more correct.

https://drafts.csswg.org/cssom/#serialize-a-css-value
> If component values can be omitted or replaced with a shorter representation without changing the meaning of the value, omit/replace them.

[Tim Nguyen (:ntim)]

Created attachment 463505 [details]
Testcase from chromium

[Tim Nguyen (:ntim)]

https://github.com/web-platform-tests/wpt/blob/a6e9e432ca/css/css-multicol/parsing/columns-valid.html should probably be extended to cover everything in the testcase.

Also I notice WebKit is inconsistent on omitting explicit or implicit `auto`.

Probably makes sense to omit in both case?

[Oriol Brufau]

I think WebKit should just remove the notion of implicit values.

[Radar WebKit Bug Importer]

<rdar://problem/102287297>

[Tim Nguyen (:ntim)]

Pull request: https://github.com/WebKit/WebKit/pull/6441

[Emilio Cobos Álvarez (:emilio)]

Yeah agreed omitting redundant values is more correct, in Gecko this is what causes the behavior: https://searchfox.org/mozilla-central/rev/219df29d0fb5d8928ae41bba4a605046de411cf0/servo/components/style/properties/shorthands/column.mako.rs#11

I'd be happy to change to follow suit here if you file a bug :)

[Ahmad Saleem]

(In reply to Emilio Cobos Álvarez (:emilio) from comment #8)
> Yeah agreed omitting redundant values is more correct, in Gecko this is what
> causes the behavior:
> https://searchfox.org/mozilla-central/rev/
> 219df29d0fb5d8928ae41bba4a605046de411cf0/servo/components/style/properties/
> shorthands/column.mako.rs#11
> 
> I'd be happy to change to follow suit here if you file a bug :)

https://bugzilla.mozilla.org/show_bug.cgi?id=1800394

[EWS]

Committed 256625@main (492c75edf7c7): <https://commits.webkit.org/256625@main>

Reviewed commits have been landed. Closing PR #6441 and removing active labels.