Re: [css3-flexbox] Computed value of "flex-item-align:auto" on the root node

On Friday, May 11, 2012, Daniel Holbert wrote:

> On 04/30/2012 03:51 PM, Tab Atkins Jr. wrote:
> >>  # Computed Value:
> >>  #  ‘auto’ computes to parent's ‘flex-align’; otherwise
> >>  #   as specified
> >>  # [...]
> >>  # A value of ‘auto’ for ‘flex-item-align’ computes to
> >>  # the value of ‘flex-align’ on the flexbox item's flexbox.
> >>
> >> http://dev.w3.org/csswg/css3-flexbox/#flex-item-align
> >>
> >> I see two ways to handle this situation:
> >> (a) if there's no parent, 'auto' computes to 'stretch'
> >>      (the default flex-align value)
> [...]
> >
> > Whoops, silly mistake.  It should definitely be (a).  I'll make the
> > change shortly - fantasai and I are hip-deep in the Pagination section
> > right now.
>
> I just wanted to follow up on this -- I don't think this
> change has made it into the spec yet.
>

I made the change earlier today, actually - I was waiting for a WG
resolution on it, which I got yesterday.  Now, ''auto'' computes to the
computed value of the main size property.

Ooh, I just remembered that it's in my stack of uncommited patches right
now.  I'll commit it tomorrow.


> Also: The two chunks of spec that I quoted above are a bit
> contradictory.  The former "Computed Value" chunk just refers to the
> "parent" & is non-flexbox-specific, whereas the prose below it refers to
> the "flexbox item's flexbox" & leaves it undefined what "auto" would
> compute to in non-flexbox contexts.
>
> I strongly prefer the more general language in the "Computed Value"
> chunk, rather than the flexbox-specific prose.
>

Luckily, the issue here with not knowing whether a child is an item or not
has been resolved - HTML's replaced elements are now *always* flexbox items
regardless. This means that whether something is a flexbox item or not is
based purely off of computed-time information, so I can go back to having
these properties refer to flexbox items.

~TJ