Re: Unprefixing intrinsic size keywords (-moz-min-content et al)

On 08/27/2015 05:49 PM, L. David Baron wrote:
> On Thursday 2015-08-06 09:42 -0700, [email protected] wrote:
>> We (blink) would like to unprefix the intrinsic sizing keywords:
>> https://drafts.csswg.org/css-sizing/#width-height-keywords
>>
>> We support them for widths and for heights (though they all do the same thing for heights)
>>
>> We were wondering what Gecko's plans for those keywords is -- are you also interested in unprefixing? Are you happy with the keywords as specced? Would you rather stop supporting them?
>
> I'm largely happy with them; they were my idea (under different
> names) originally, although that proposal predated serious work on
> vertical writing modes, which makes things a bit more complicated.
> I haven't looked closely at the relevant specs lately, but I think
> the controversial spec parts are the actual definitions of intrinsic
> sizes, whereas the spec for the keywords is pretty trivial.  Though
> at a quick glance I think the spec at
> https://drafts.csswg.org/css-sizing/#width-height-keywords for fill,
> max-content, min-content, and fit-content is fine, at least for the
> inline-size direction.  (This of course requires that the actual
> definitions exclude the effects of the (min-/max-/)(width/height)
> properties, which I believe they do.)
>
> My biggest concern is probably what they mean when used in the block
> direction.  I'm not sure we currently do anything sensible there, or
> that the spec has sensible definitions for max-content, min-content,
> and fill-available sizes there.

The CSSWG resolved (possibly re-resolved? I couldn't find the original
minutes, though I recall discussing the issue) on this here:
   https://lists.w3.org/Archives/Public/www-style/2015Aug/0109.html
I think 'fit-content' can be safely thrown in with min/max-content.

For 'fill', I'm less sure we're solid.

A quick test strategy would be:
   1. Fill a box wit content and give it properties P
   2. For the min-content test
        * assign it 'width: min-content' in the test
        * assign it a zero-width container in the reference
   3. For the max-content test
        * assign it 'width: max-content' in the test
        * assign it maximally large container in the reference
   4. For the fit-content test
        * assign it 'width: fit-content' in the test
        * leave its container as auto
   5. Make the reference boxes all 'float: left' or something else
      to give them shrinkwrapping behavior.
as long as the test and reference match for all content C and all
properties P, the implementation of the keywords is correct, even
if the layout algorithms aren't quite there yet. :)

~fantasai