[css-contain] Size contain shouldn't apply to tables

[MReschenberg]

The section on size containment says size containment has no effect on internal table elements, however it seems like it also shouldn't apply to tables (which aren't internal table elements).

cc: @dbaron

[dbaron]

Yeah, I don't think it makes sense for it to apply to tables, because tables don't have a sensible way of being smaller than their contents. That's never been defined and it's not particularly clear what their contents would do (i.e., how wide to make each column).

[frivoal]

Oops, that sounds right, I don't quite see how it would work either. Sorry I phrased that wrong.

Ping: @tabatkins @eaenet @mrego. I suspect you all agree, but just wanted to check before updating the spec.

[tabatkins]

Yeah, sounds reasonable

[frivoal]

Agenda+ to confirm the proposed change.

[jonjohnjohnson]

I can recall times when I've used tables with table-layout: fixed and width not set to auto, where I'd be surprised to find out contain has no effect. Would this then force me to use a bunch of non-semantic divs with table structure display values? I'm primarily thinking of highly responsive layouts where I'm simulating element query behaviors, guessing I'd gain decent perf from using contain.

[frivoal]

With table-layout:fixed, you can set the width of the table to less than its min-content width, so that sounds tempting, but:

• you still cannot set the height to less than the min-content height
• what happens when the width is less than the min-content is not fully interoperable

So, I think we still need to exclude tables entirely, including when table-layout is fixed.

[tabatkins]

Would this then force me to use a bunch of non-semantic divs with table structure display values?

Note that when we say "tables" here, we're referring to anything using display:table. So switching to <div style="display:table"> won't change anything.

[jonjohnjohnson]

Note that when we say "tables" here, we're referring to anything using display:table.

Ya, I figured. :)

[mrego]

Sounds good to me too.

[css-meeting-bot]

The Working Group just discussed Size contain shouldn't apply to tables, and agreed to the following:

• RESOLVED: size containment doesn't apply to tables

The full IRC log of that discussion

<melanierichards> topic: Size contain shouldn't apply to tables
<melanierichards> Github: https://github.com//issues/2746
<melanierichards> florian: we've already said that size containment doesn't apply to table parts, haven't said anything about tables. the way containment works, tables would end up smaller than their contents
<melanierichards> Florian: in the inline axis of tables we don't have a good definition or interop. In block axis seems to do nothing. In inline axis, in some browsers the borders go diagonal
<melanierichards> Florian: think we should skip this here
<melanierichards> fantasai: I can imagine someone wanting to size constrain a table and put a scroller on it
<melanierichards> Florian: proposal is to add tables to the list of things size containment doesn't apply to
<melanierichards> Rossen: sounds reasonable
<melanierichards> Rossen: objections? opinions?
<melanierichards> [none]
<melanierichards> RESOLVED: size containment doesn't apply to tables

[mrego]

BTW, there's already a test currently checking the opposite thing: http://w3c-test.org/css/css-contain/contain-size-012.html

I'll update the Chromium implementation and modify it.

[mrego]

The test has been modified in web-platform-tests/wpt#11502.