[css-cascade-6] What is the desired shape of CSSScopeRule?

[mirisuzanne]

In #8468 @andruud pushed a spec definition for adding CSSScopeRule to CSSOM:

	[Exposed=Window]
	interface CSSScopeRule : CSSGroupingRule {
	  readonly attribute CSSOMString start;
	  readonly attribute CSSOMString end;
	};

As far as I can tell, this change in the spec never had a resolution attached – so bringing it up here to get that resolution. In conversation with @fantasai, she also raised a few questions about the final shape of this rule:

• Is 'start'/'end' the terms we want? Or do we want 'root'/'floor', or any number of other things. Since this is the only place in which the terminology is actually user-facing, it becomes a real question.
• Should they return null string or null if they're missing?
• Do they include the wrapping parentheses or not?

[fantasai]

Fwiw, I think we should not include the wrapping parentheses so that the returned string is a proper selector, and they should be null if omitted.

I don't have much opinion on the names (they seem fine), but I think it's a question worth asking.

[mirisuzanne]

I agree that the names seem fine to me. Other terms that get used for these concepts are 'root', and 'boundary' or 'limit'.

[tabatkins]

I'm fine with the names as well. I could think of some different ones, but nothing seems better, and quite a bit seem worse.

null when omitted is indeed correct imo. There's a lot of stuff in CSSOM that uses empty string for this semantic and I think that's just terrible legacy design, and doesn't need to be emulated.

Yeah, omit the wrapping parens so the values are just the selectors themselves. The parens are a syntactic necessity for the rule itself, they're not part of the start/end bits.

[tabatkins]

Well, wrt the names, I actually suggested upper and lower as option-bag property names in qSA() over in #7709 (comment). Those might be a little better?

[bramus]

Does upper and lower make sense with @scope-siblings? My hunch would be that start and end is better, especially if the suggested :scope-end selector would make it.

[tabatkins]

Yeah that's a good point, so let's stick with the more generic start and end.

[mirisuzanne]

Does this seem like a good candidate for async resolution? I think so. Proposal is:

• Keep start/end terms
• Return null when missing
• Don't include parenthesis

[bramus]

SGTM

[css-meeting-bot]

The CSS Working Group just discussed [css-cascade-6] What is the desired shape of CSSScopeRule?, and agreed to the following:

• RESOLVED: Accept the CSSScopeRule design, with the three changes noted in the thread.

The full IRC log of that discussion

<TabAtkins> miriam: Anders pushed a definition for the CSSSCopeRule, in the ED
<TabAtkins> miriam: We didn't ahve a resolution on the details tho, so wanted to clarify it's defined properly
<TabAtkins> miriam: Main qs in the thread are terms used for the start and end fo the scope (sometimes called "root" and "lower boudnary"/"scoping limit")
<TabAtkins> miriam: In thread we decided start/end were clearest terms, especially if we do sibling scopes - they keep making sense, while upper/lower dont'
<TabAtkins> miriam: Another q is what to return if one of the values isn't specified, we suggest null
<TabAtkins> miriam: Final q, should the value contain the parens around the selectors? Propose no.
<Rossen_> q?
<TabAtkins> miriam: So proposal is accept the CSSScopeRule, with the three conclusions outlined above.
<fantasai> +1
<bramus> +1
<TabAtkins> Rossen_: Looks like agreement in the thread, anyone have objections?
<TabAtkins> RESOLVED: Accept the CSSScopeRule design, with the three changes noted in the thread.