On Wed, Apr 25, 2012 at 8:29 AM, Alex Mogilevsky <alexmog@...> wrote:
> ± From: fantasai [mailto:fantasai.lists@...]
> ± Sent: Friday, April 20, 2012 4:03 PM
> ± There are several places in the Flexbox algorithm where it seems assumed that
> ± the writing mode of the flexbox item is the same as that of the flexbox. This
> ± isn't a true assumption, so the algorithm needs to be fixed to define these
> ± cases.
> Can you point at such places? Flexbox is generally agnostic to what happens inside the items, including their inner layout type and most certainly writing mode. Issues of mixed writing modes have to be addressed in writing-modes spec or in specs for individual layout types (for issues of being inside a parent with a different writing mode)...
> If there are still places where flexbox spec says something that implies same writing mode in a child, it must be corrected.
> Perhaps you refer to baselines? Column-direction flexbox always treats 'baseline' alignment as 'center', while its children can have baselines in block direction. We should fix that.
> What else?
I believe fantasai and I have fixed all of the places where this
occurred. It was mostly just in the "hypothetical main size of items"
step, because that ends up invoking some subtle behavior with a lot of