[css-grid] Don't expand flexible tracks under a min-content constraint

[Loirooriol]

Consider this code: https://jsfiddle.net/wjthokzf/

#wrapper {
  width: 0;
}
#grid {
  display: grid;
  float: left;
  grid-template-columns: minmax(0, 1fr);
  border: solid;
}
#item::before, #item::after {
  content: '';
  float: left;
  height: 50px;
  width: 50px;
}

<div id="wrapper">
  <div id="grid">
    <div id="item">div>
  div>
div>

The grid is sized under a min-content constraint. The sizing algorithm reaches https://drafts.csswg.org/css-grid/#algo-flex-tracks with a base size and growth limit of 0 for the column.

The free space is indefinite, so the used flex fraction is the result of finding the size of an fr, called with the column and the max-content contribution of the item (100px) as the space to fill. The leftover space is 100px, the flex factor sum is 1, the hypothetical fr size is 100px, and this is what the algorithm returns. So the flex fraction is 100px. And the base size of the column is set to 100px.

However, neither Chromium, Firefox nor Edge do this. The column is 0px wide.

So I think that this "Expand Flexible Tracks" step should only happen if the available space is definite, or if the grid is being sized under a max-content constraint. But not when under a min-content constraint.

[fantasai]

Fixed to match implementations. I wonder why we did this for Maximize Tracks but not Expand Flexible Tracks. Agenda+ to request review, esp from @atanassov and @tabatkins. r=Oriol

[css-meeting-bot]

The CSS Working Group just discussed Don't expand flexible tracks under a min-content constraint, and agreed to the following:

• RESOLVED: Accept proposal in https://github.com/w3c/csswg-drafts/issues/3683

The full IRC log of that discussion

Topic: Don't expand flexible tracks under a min-content constraint
github: https://github.com//issues/3683
fantasai: Case where spec forgot to consider min content correctly. Impl do logical and don't expand track to take up space. hcanging spec to match impl and do thing you expect which is size to smaller end when under min-content constraint.
fantasai: If you can shrink something down without overflow then min-content constraint should be that amount and not bigger. spec violated concept, impl did correct. Trying to match them up
astearns: Any comment? I not you asked for TabAtkins or Rossen_ to comment
fantasai: I'd prefer to get their +
TabAtkins: I'll review shortly
astearns: Resolve or wait?
TabAtkins: I trust fantasai so resolve. If I find a mistake I'll say something
Rossen_: On the same page. Proposed doesn't seem crazy, just need to look at overall algo fit. I'm sure fantasai spent more cycles so I trust her
astearns: Other comments?
astearns: Objections to this change?
RESOLVED: Accept proposal in https://github.com//issues/3683
astearns: All three of these look like they need tests or need tests verified. Have ther ebeen any?
fantasai: None in wpt yet. I'll check with oriol and he might know more
astearns: TabAtkins as you review can you check in tests?
TabAtkins: Sounds good
astearns: Once we have tests anything to keep us from updating CR?
fantasai: Prob tests for other things. I think most that should be fixed is but there might be one or two not.
astearns: I suspect no DoC yet.
fantasai: Right. Bulk of work is that and changes section
astearns: Anything else on grid?
fantasai: I'm going to say no