[css-values-4] Punt attr() and toggle() to Level 5

[fantasai]

Tab and I think it's time to start preparing css-values-4 for CR, as a lot of the stuff in here is already implemented (and shipped!). We think it would make sense, however, to cut ''attr()'' and ''toggle()'' and push them to Level 5. Agenda+ to ask about doing this.

[Loirooriol]

Given #6764, toggle() doesn't seem ready for CR. I support deferring it to L5.

[Loirooriol]

Should mix() also be deferred to L5?

[fantasai]

@Loirooriol No, we need that one to handle serialization of interpolation of transforms.

[css-meeting-bot]

The CSS Working Group just discussed #6753, and agreed to the following:

• RESOLVED: Punt attr() and toggle() to Level 5

The full IRC log of that discussion

<dael_> Topic: #6753
<fantasai> github: https://github.com//issues/6753
<dael_> fantasai: We think we should drop from L4. attr() and toggle(). not quite solid. A bunch of the rest of the spec has been impl
<dael_> fantasai: Narrowing spec to prep for CR
<dael_> Rossen_: Obj?
<dael_> oriol: Question on this. I wonder if mix should also be deferred. Added recently and issues about syntax
<dael_> fantasai: I think problem with mix is need it to serialize interpolation of certain transforms. I think we should keep it and work out the issues
<dael_> oriol: Okay
<dael_> Rossen_: Hearing no objections
<dael_> RESOLVED: Punt attr() and toggle() to Level 5
<fantasai> oriol, we kept it simple though, and are pushing all the more complex interpolations to L5

[SelenIT]

Given that attr() has been there since at least 2005 and keeps being deferred because of no implementation signal, shouldn’t it be reconsidered more deeply? The existing situation seems much like the lack of implementers’ interest. Perhaps the attr() function should forever stay in the string-only form as defined in CSS2.x, and for deriving typed values from attributes something new should be introduced instead, to avoid confusion and make experimenting/implementation easier?

[Crissov]

I don’t really disagree, but this idea was basically rejected in #3702.

[SebastianZ]

The question is how to get implementer interest because author interest is definitely there regarding its many use cases.

Sebastian