[css-grid] span=1 and span>1 use different size contribution to 'auto' min-sizing

[MatsPalmgren]

https://drafts.csswg.org/css-grid/#algo-content
Step 1 : "For auto minimums" says:
If the track has an auto min track sizing function and the grid container is being sized under a min/max-content constraint, set the track’s base size to the maximum of its items’ min/max-content contributions, respectively.

In Step 2 : "For intrinsic minimums" there is no such special handling for 'auto' min-sizing.

It seems a bit odd that span=1 and span>1 items should contribute different sizes to 'auto' tracks.
Is this an oversight and that the above paragraph also applies to span>1 items for 'auto' min-sizing tracks?

[tabatkins]

I think this is handled by steps 2.2 and 2.3 there? We apparently thought it was better to split the auto-handling separately for the two cases in step 2, while step 1 handles it in one spot.

[fantasai]

Marking invalid; please re-open if this is incorrect. :)

[MatsPalmgren]

2.2 is labeled "For content-based minimums" which to me means
'min/max-content', so I find the label misleading (because the spec
text under it now also includes 'auto').

2.3 is labeled "For max-content minimums" which to me means
'max-content', so I find the label misleading.

Just to make sure I understand the span>1 handling of 'auto' correctly:
2.1 "For intrinsic minimums", clearly includes 'auto' (unconditionally),
so this means that an item that spans an 'auto' track first contributes
its min-size contribution under 2.1, then also (one of) its min/max-content
contribution under 2.2/2.3 (under intrinsic sizing).
So the net effect for 'auto' in 2.1 - 2.3 is slightly different from
the span=1 case due to the "Otherwise" there, is that intentional?

[MatsPalmgren]

After thinking some more about the 'auto' case, I suspect that applying
both its min-sizing (2.1) and min/max-content (2.2/2.3) contributions
is probably correct for the span>1 case. We apply both contributions
for 'min-content' and 'max-content' too in the span>1 case, but not
in the span=1 case. So if the goal of these changes were "make 'auto'
min track sizes behave as 'min(max)-content' when sizing under
a min(max)-content constraint" then the current text looks correct.

If so, I think it would be equivalent to stating upfront that "if sizing under
a min(max)-content constraint then treat an 'auto' min track sizing
as a 'min(max)-content' track min sizing in the following rules"
(for any span).

[fantasai]

I don't think there's an intentional difference... what is the difference you're noticing?

[MatsPalmgren]

I don't think I can explain it better than I did in the last comment.

Anyway, the problem I originally reported doesn't exist so I'm fine with
the current spec text in that regard.

[MatsPalmgren]

You might want to fix the editorial issue with the labels I mentioned above though...

[fantasai]

Tab and I looked over this again and decided to handle it as you suggested in #500 (comment) ... in particular as 2.2 seems to have left out max-content constraint handling for auto, even though max-content is handled there, and that seems like an error.

[fantasai]

Fixed in d8a33b9 . Please let us know if this seems correct now. :)