[css-flexbox] Should a flex container's "overflow directions" be established by its main/cross axes? (instead of its block/inline directions)

Hi www-style,

QUESTION:
When content overflows off the top of a flex container with
"flex-direction:column-reverse;overflow:auto", should scrollbars appear?

As a testcase, see the right half of:
 https://bugzilla.mozilla.org/attachment.cgi?id=8460329

* Chrome 36 does render a scrollbar on the right.
* Firefox 30 and IE 11 do not render a scrollbar on the right.

I think the Chrome behavior probably makes more sense, but I think
there's some spec-smithing that would be needed to make this actually
correct.

SPEC REFERENCES:
The CSSOM-View "scrolling area" table seems to define what parts of a
scrollable region are reachable (e.g. whether the scrollbar stops at the
container's own padding-top edge, vs. the top of the topmost descendant):
 http://www.w3.org/TR/cssom-view/#scrolling-area

That table relies on the "overflow directions", which are defined here
as depending on the *block flow direction & the inline base direction*:
 http://www.w3.org/TR/cssom-view/#overflow-directions

I think this definition of "overflow directions" needs expanding. For a
flex container (and for a grid), it seems to me that it's not useful to
rely on the block direction & inline direction here -- those aren't
relevant for predicting where things will overflow as content is added,
and where the author will expect things to be scrollable.

Instead, I'd expect that the *main axis* and *cross axis* should be used
for determining the "overflow directions", for a flex container & grid
container.

Thoughts?

Thanks,
~Daniel

P.S. [Apologies if 2 copies of this message end up on the list; I
initially sent this from the wrong email account, & I just got the
"awaiting moderation" bounce message. I'm re-sending from my @mozilla
address, which is the one I'm actually subscribed under.]

Received on Tuesday, 22 July 2014 18:49:37 UTC