[css-flexbox] when fragmenting a column-reverse flex container, which page gets the first flex item?

tl;dr: When a UA paginates (fragments) a "column-reverse" flex
container, should the first flex item end up at the bottom of the *last*
page or the bottom of the *first* page?

EXAMPLE:

Consider this HTML markup:
 <div style="display: flex; flex-direction: column-reverse">
   <div>A</div>
   <div>B</div>
   <div>C</div>
   <div>D</div>
   <div>E</div>
 </div>

So, the on-screen rendering looks like this:
  +-------+
  |   E   |
  |   D   |
  |   C   |
  |   B   |
  |   A   |
  +-------+

QUESTION: If we paginate this content, where should the first flex-item
("A") appear, when printed?

Option 1: first item goes at the bottom of the *last* page:
   Pg1    Pg2    Pg3
  +---+  +---+  +---+
  | E |  | C |  | A |
  | D |  | B |  |   |
  +---+  +---+  +---+

Option 2: first item goes at the bottom of the *first* page:
   Pg1    Pg2    Pg3
  +---+  +---+  +---+
  | B |  | D |  | E |
  | A |  | C |  |   |
  +---+  +---+  +---+

Right now Chrome implements Option 1.  However, I think I prefer Option
2, and I also believe the spec's sample fragmentation algorithm yields
Option 2 (if you read the heading "Single-line column flex container" to
include both column and column-reverse, which is reasonable IMHO given
that there's no sample algorithm at all provided for column-reverse.)

In particular, the spec's sample algorithm for "Single-line column flex
container" says:
  # 2. Lay out as many consecutive flex items... as possible...
  # starting from the first, until there is no more room on the
  # page or a forced break is encountered.
http://dev.w3.org/csswg/css-flexbox/#pagination-algo

So that pretty clearly indicates that the first flex item should end up
on the first page.

Moreover, Option 2 is very analogous to how multi-line flex containers
partition their items into flex lines. (The first flex item always ends
up on the first flex line, regardless of whether we're column or
column-reverse.)

Thoughts? Perhaps the spec could be clarified on this point? (If there's
general agreement that Option 2 is reasonable, then maybe we should
change the sample-algorithm header-text "Single-line column flex
container" to explicitly mention "column-reverse", and similar for the
other sample algorithms?)

Thanks,
~Daniel

Received on Tuesday, 23 April 2013 22:00:25 UTC