Should we push for equivalent/interoperable support for popover in MathML and SVG?

[bkardell]

Recently, HTML has added 'popovers' - which allow some of the machinery related to various kinds of popup ui to be natively supported. It is getting widely implemented now, but is still of a bit limited use without its corresponding partner CSS Anchor Positioning. Once we have that too, I expect use to really take off as it makes it easy to do lots of useful popover/popup stuff.

However, while he originally had intended it for everywhere, @mfreed7 was encouraged to focus on HTML first at least. The result is that that stuff is landed without similar support in MathML or SVG. At least for MathML-Core this seems like it would be very useful and worth moving up to HTMLOrForeignElement or something?

@AmeliaBR @nikolaszimmermann

[mfreed7]

Thanks for raising the issue. I am (and was) supportive of supporting popover on Element rather than HTMLElement. @domenic was, I believe, the one that requested it be restricted, but I've since forgotten (and can't find) the rationale. Perhaps he can comment here, in case there's still a good reason not to support it.

[domenic]

I think we discussed this in whatwg/html#8221 (comment) a bit. The precedent of hidden="" only applying to HTML is what tipped us over the edge to starting with only HTML.

Some thoughts:

• Much functionality is HTML-specific today, e.g. most of https://html.spec.whatwg.org/#global-attributes . This problem might be better tackled holistically. (Also, as noted below, there's some tech debt to clean up before proceeding, I think.)
• There's a general choice between "all namespaces" (Element) and "HTML + MathML + SVG" (HTMLForeignElement). Different features have made different choices here.
  • E.g., id="", class="", and slot="" are "all namespace" functionality.
  • Various parts of https://html.spec.whatwg.org/#htmlorsvgelement are, I think, supposed to be expanded to the main three. But, nobody ever finished that spec work, or more specifically, nobody finished that test work so we weren't able to merge the spec work: Rename HTMLOrSVGElement to reflect its wider use in MathML as well whatwg/html#5248 .
• UA stylesheets are currently namespace-specific. [css-cascade-4] An "all-namespaces" user agent stylesheet csswg-drafts#8573 would need to be resolved before we had a proper place to put things that applied in all namespaces.
• You might also want to distinguish between "works with other-namespace elements" vs. "works in other-namespace documents". Is it a good idea to give pure SVG documents (which some might think of as images) the ability to create popovers? Or MathML documents?
• It's unclear whether you want to port over popovertarget="" and popovertargetaction="" as well.

[mfreed7]

Yep, those are indeed some good reasons. It sounds like nothing is really "fundamental" but more that there is a lot of clarification and cleanup needed before we could do this. Sounds reasonable to me.

• You might also want to distinguish between "works with other-namespace elements" vs. "works in other-namespace documents". Is it a good idea to give pure SVG documents (which some might think of as images) the ability to create popovers? Or MathML documents?

This is a great question. My gut reaction is that we want the former (works with other namespace elements) and not the latter. It seems weird to have part of an SVG be a popover. Same for mathml. Though I suppose I could imagine some cool effects where part of an image/equation animates up and over to emphasize it?

• It's unclear whether you want to port over popovertarget="" and popovertargetaction="" as well.

Since AFAIK there aren't button-like things in SVG or math-ml, I think I'd say no?

[bkardell]

since it was pointed out that we could use examples/info about the API

https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/popover
https://open-ui.org/components/popover.research.explainer/