Csswg-drafts: [css-cascade-3] Publish Cascading and Inheritance 3 as a REC

[xfq]

We should close the only open issue, #5313 (there's a PR for it: #5358). In the test suite, 96 of 105 required tests meet CR exit criteria.

There doesn't seem to be any non-trivial change since the last CR.

[fantasai]

96/105 is not 100%. Has anyone analyzed the failures to see why they're failing? Also, do we actually have coverage of the spec? Number of tests isn't an indicator of coverage, it's just a number that doesn't mean very much.

[frivoal]

A fair few of the missing tests involve things like "load this file as the UA style sheet", or "select the alternative stylesheet", which aren't things that browsers are required to be able to do. So maybe they're not quite test failures. But nonetheless, we would indeed need a coverage analysis and an actual look into what tests are failing / missing and why to see if we're good to go or not.

[fantasai]

Implementation Report: https://drafts.csswg.org/css-cascade-3/implementation-report

[fantasai]

I added a few tests. Assuming the CSS2 test suite has full coverage of its own features, we now have full coverage of Level 3.

[svgeesus]

A fair few of the missing tests involve things like "load this file as the UA style sheet", or "select the alternative stylesheet", which aren't things that browsers are required to be able to do.

Those should be marked as optional tests in that case. IIRC CSS2 required this, so it has already been tested in the past.

[frivoal]

I think the behavior to be followed if there is a different file as the UA style sheet, or if the UA let's the user selects an alternative stylesheet is required. However, whether or not these features exist in the first place is largely a matter of UX, and we cannot really required anything.

Regardless, AFAIK, the spec has not changed with respect to these things from CSS2, and CSS2 is a REC already, so we're fine here.

@fantasai's report shows that we're clean on things that are new to level 3, so all in all, I think we're good to go.

[svgeesus]

Agreed with Agenda+ to discuss transition to Proposed Recommendation.
Assuming that succeeds, who will open the issue on the transitions repo?

[css-meeting-bot]

The CSS Working Group just discussed [css-cascade-3] Publish Cascading and Inheritance 3 as a REC, and agreed to the following:

• RESOLVED: Move forward with Cascade L3 REC without CSS 2 tests

The full IRC log of that discussion

<dael> Topic: [css-cascade-3] Publish Cascading and Inheritance 3 as a REC

<dael> github: https://github.com/w3c/csswg-drafts/issues/5396

<dael> Rossen_: It'll transition to PR.

<dael> Rossen_: Do we have florian chris or fantasai on?

<dael> fantasai: I'm here

<dael> Rossen_: What is our readiness and what do we need to consider?

<fantasai> https://drafts.csswg.org/css-cascade-3/implementation-report

<dael> fantasai: Have impl report to everything new to L3^

<dael> fantasai: 2 passes

<dael> fantasai: Went through test suite, added so we have full coverage for new parts

<dael> fantasai: Do we want to transition to PR or do we need to cover parts of spec that are CSS 2 since that's a lot more work

<dael> Rossen_: Opinions?

<dael> Rossen_: Anyone who thinks we should cover parts that are css 2?

<dael> Rossen_: Not hearing any desire expressed

<dael> xfq: We can link to css 2 directly in the report and add css 2 tests in the implementation report

<dael> Rossen_: WOuld that be okay fantasai ? Is that straightforward to link ot existing test results

<dael> fantasai: impl that went through css 2 are quite old and not the same as the ones that passed L3. There would be 2 passes, but not the same. I also don't know where css2 impl report is. Let's see if I can find it

<fantasai> http://test.csswg.org/suites/css21_dev/20110323/report/

<dael> fantasai: Looks like this is css 2 test suite ^

<dael> astearns: If there were up to date from wpt that would be one thing, but we don't have css 2 suite in wpt

<dael> fantasai: They're in but require manual configuration so can't be automated by wpt. Looking at wpt would give fails that are not actual fails

<dael> astearns: You're saying it's a bunch of additional work to retest this spec coverage in css 2 tests

<dael> fantasai: It would be...it won't take a lot of time but probably a day if there are instructions on how to load the user stylesheet for the implementations

<xfq> https://www.w3.org/Style/CSS/Test/CSS2.1/20110323/reports/

<dael> Rossen_: I'm more interested in seeing how to move it without css2 retesting. Is there a reason we shouldn't? I get that this is a can we do it, but why?

<fantasai> test results for CSS2 - http://test.csswg.org/suites/css21_dev/20110323/report/results.html

<dael> Rossen_: Can we resolve without css 2 work?

<dael> Rossen_: Any objections to that?

<dael> Rossen_: Prop: Move forward with Cascade L3 REC without CSS 2 tests

<dael> Rossen_: Objections or reasons why we shouldn't do it?

<fantasai> s/without/without retesting/

<dael> RESOLVED: Move forward with Cascade L3 REC without CSS 2 tests

<dael> Rossen_: Who will handle? fantasai or florian ?

<dael> fantasai: Me

[svgeesus]

Proposed Rec requested 9 Sept 2020, approved 18 Dec 2020

Publication as PR requested 21 Dec, expected 22 Dec