[css-scrollbars] Upgrade to Recommendation?

[lukewarlow]

Is the CSS Scrollbars Level 1 specification ready to published as a recommendation? Currently it's a candidate and has been for a while. The only changes from the current draft compared to the CR is a minor correction.

It has long been supported in Firefox, but will soon also be supported in Chromium (see my I2S), it also has a positive standards position from WebKit scrollbar-width position scrollbar-color position.

[svgeesus]

Most of the 115 tests are passed and there is only 1 open issue. There have been only minor changes since CR.

@tantek @frivoal @atanassov WDYT?

[lukewarlow]

There's also this #9169 but it doesn't block shipping imo was more just an observation I had.

Also just fyi the 1 test failing in Chrome on wpt.fyi is fixed in Canary

[svgeesus]

• scrollbar-color Interop 2024 proposal
• scrollbar-width Interop 2024 proposal

WPT results show full interop of 115/115 for Chrome (and Edge) and 115/115 for Firefox. Safari trails somewhat with 74/115

My +1 for requesting Proposed Recommendation

[css-meeting-bot]

The CSS Working Group just discussed [css-scrollbars] Upgrade to Recommendation?.

The full IRC log of that discussion

<florian> q+
<Rossen_> ack florian
<Frances> Chris L: css scrollbars have not changed much. 115/115 for chrome, edge, and Firefox, safari a little bit less. Accessibility concerns are scroll-bars do not need AA and scrollbars with people in cognitive disability. Both flagged.
<Frances> Florian: Some tweaks to the bike shed, some notes that we changed to draw a diagram. Should draw a diagram, we have a bunch of tests, but I think they are incomplete. Some tests are not according to the spec. Testing specified and out of scope elements.
<Frances> Florian: If you assume the feature works at all, does it work correctly in the test? Very few tests do it correctly currently.
<PaulG> q+
<Frances> Florian: If make track transparent, scrollbar might have up and down buttons or the spec might use color only and do a gradient if rounded or shaded. Some tests assume wrongly behaviors and implementations.
<fantasai> s/If you assume the feature works at all, does it work correctly in the test?/The tests assume the feature works at all, and if that's true, test other assumptions. But there's no test for whether the feature works at all./
<Frances> Florian: Not an issue with the spec, it is an issue with the site. Authors should not do this. Just because there is white text on a white background, would be bad to test possibly since it is not realistic.
<emilio> q+
<Frances> Paul: When a developer alters a user-agent component take on the onus of passing from most auditors, such as changing colors from 3 to 1, it would fail. If we don't accommodate 3:1 the target size would not work with auto. Transparent is a visual affordance.
<florian> q+
<Rossen_> ack PaulG
<Rossen_> ack emilio
<chris> sorry, promised this was a quick issue or take it back to github.
<Frances> Emilio: It is a good way to provide a scrollbar:hover.
<Frances> Florian: spec already contains a fair amount. If aspects that are missing, would be welcome to adding them.
<Frances> GitHub-bot #5426

[frivoal]

As a result of reading the spec and test suite to check if they were indeed ready for REC, I have a found a few issues, and filed a few spec bugs as well as a few test bugs accordingly.

• spec issues: css-scrollbars-1 Current Work

• test issues: https://github.com/web-platform-tests/wpt/labels/css-scrollbars

We should try to reduce these to zero before pushing as PR.

Help welcome as well for making PRs that would solve the issues inline in the spec (they're about making diagrams and illustrations).

• https://drafts.csswg.org/css-scrollbars-1/#issue-94d93028
• https://drafts.csswg.org/css-scrollbars-1/#issue-012ab64f

See also https://drafts.csswg.org/css-scrollbars-1/test-coverage for some information about the overall coverage of this spec.

All in all, I don't think we're that far off, but we're not there yet.

---

Side note:

Safari trails somewhat with 74/115

This says more about the quality of the test suite than about the status of interoperability, given that as of now, Safari does not implement either of the properties defined in the specification.