[Flexbox] Webkit picks a different flex base size for multicol elements with "align-self:stretch" vs. other "align-self" values

• Component: Layout and Rendering
• Status: NEW ---

[Daniel Holbert]

STR:
Load testcase, which has two auto-sized multicol elements as flex items in a vertical flex container.
 NOTE: The multicol elements are identical except for in their "align-self" values.
   - the orange one has "align-self: stretch;" (default)
   - the purple one has "align-self: flex-end;"

EXPECTED RESULTS: They should have the same height, since they only differ on "align-self", and the flexbox spec has us resolve the main-size (the height, in this case) before we take "align-self" into consideration.

ACTUAL RESULTS: Their heights are different.

For reference, see this www-style thread where I asked for clarification on this:
  http://lists.w3.org/Archives/Public/www-style/2012Oct/0017.html

Tab and I agree that Webkit renders the purple flex item correctly, but the orange one incorrectly. (it should be as tall as the purple one)

Tested using google chrome dev channel:
Version 23.0.1271.10 dev

[Daniel Holbert]

(In reply to comment #0)
> Tab and I agree that Webkit renders the purple flex item correctly, but the orange one incorrectly. (it should be as tall as the purple one)

citation: http://lists.w3.org/Archives/Public/www-style/2012Oct/0053.html

[Ojan Vafai]

I expect we have lots of multicol/flexbox bugs. We really haven't done any testing of the intersection of the two. :(

[Tab Atkins]

This intersection is sketchy anyway, since officially "width: max-content" isn't defined for multicols yet (we'll define it at <http://dev.w3.org/csswg/css3-sizing/#multicol-intrinsic> eventually).

But our behavior is inconsistent, which is incorrect regardless of the eventual definition of max-content.  (However, it appears that Firefox's behavior for max-content is the best one.)