Re: [css3-flexbox] spec contradicts itself about display: table-row; flex items

On Thu, Jul 18, 2013 at 8:03 PM, fantasai <fantasai.lists@inkedblade.net> wrote:
> I'm similarly unconvinced. It certainly seems simpler just to
> do the computed value fixup, and would probably be useful in
> cases where you want flexbox, but use table layout as a backup.
>
> .flexbox { display: table; display: flex; }
> .flexbox * { display: table-cell; /* horizontal */
>              display: table-row;  /* vertical */ }
>
> I've pushed a clarification here, but maybe we should instead
> change that part of the spec in favor of computed value fixup.
>
> bz, dholbert, Rossen, any thoughts?

We already have useful fallback behavior for floats.  I'm unconvinced
that we need to care about more modes of fallback.

The box-model fixup stuff is meant for more than just tables, though
that's the only thing that matters *so far*.  For example, ruby markup
is supposed to let you do a naked "<rt>foo<rb>bar" and have that fixed
up with a display:ruby container around it, I think.  Similarly with
run-in when we define that - we definitely want run-ins to move around
in the box tree before they get forced into being flex items.

~TJ