min-content/max-content computation WAS: [css3-flexbox] Does "width" property on a flex-item affect its used value of "min-width:auto"?

Ugh. I spoke too soon. I went to fix this and realized that this is a more
general problem than just the flexbox issue. There are a couple questions
that are not clearly spelled out in the spec when you specify both
min-width and width.

<style>
.container {
    min-width: min-content;
    width: 100px;
    height: 40px;
    background-color: blue;
}
.child {
    display:inline-block;
    height: 20px;
    width: 200px;
    background-color: pink;
}
</style>

<div class=container>
  <div class=child></div>
</div>

<div class=container>
  <div class=child>aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa</div>
</div>

1. Should container always always be 100px or should min-content override
it?
2. Should the min-content on the container use the child's min-content or
it's 200px width?

Testcase: https://bug-106143-attachments.webkit.org/attachment.cgi?id=181530
Screenshot in IE9, FF17 and Chrome tip of tree:
https://bug-106143-attachments.webkit.org/attachment.cgi?id=181529

I can see valid arguments for either of the three existing behaviors.

On Wed, Sep 26, 2012 at 9:42 PM, Daniel Holbert <dholbert@mozilla.com>wrote:

> OK, thanks for the sanity-check!
>
> ~Daniel
>
> On 09/26/2012 09:38 PM, Ojan Vafai wrote:
> > You're interpretation is correct. We just have a bug.
> >
> >
> > On Wed, Sep 26, 2012 at 9:08 PM, Daniel Holbert <dholbert@mozilla.com
> >wrote:
> >
> >> Hi www-style,
> >>
> >> Suppose you have this simple flex-item in a horizontal flex container:
> >>   <div>abc</div>
> >> with no fancy style rules affecting it.
> >>
> >> So it has "min-width:auto" (the default), and that resolves to the
> >> min-content width -- in this case, the width of the text "abc".
> >>
> >> NOW -- suppose we set "width" on the div, e.g.:
> >>  <div style="width:1000px">abc</div>
> >>
> >> MY QUESTION: Does the "width" property there affect the used value of
> >> "min-width: auto" that we pass to the flex algorithm?
> >>
> >> My interpretation: I don't think it should -- to me, the min-content
> >> width is a function of the div's contents, and it necessarily doesn't
> >> depend on the div's "width" property. (Otherwise, it seems that "width:
> >> min-content" would be problematic to define).
> >>
> >> I'm asking because Webkit's current implementation[1] seems to disagree
> >> with me on this (they render the above flex-item as if it had
> >> "min-width: 1000px"), and I want to sanity-check my own understanding.
> >>
> >> (For reference, I filed https://bugs.webkit.org/show_bug.cgi?id=97747on
> >> this, but that may or may not be a valid bug depending on the outcome of
> >> this thread.)
> >>
> >> Thanks,
> >> ~Daniel
> >> [1] I tested Chrome dev channel, ver. "23.0.1271.6 dev", Linux x86_6
> >>
> >>
> >
>