[css3-regions][css3-gcpm] Plan A versus Plan B

As far as regions are concerned, my opinion is that we should pursue both
Plan A (css3-regions) and Plan B (column selector styling). It will be
terrifically useful to be able to style individual columns, but there are
some layouts where placing regions directly will be much more
straightforward.

Multicolumn is structured to begin with (same-width, same-gutter columns
placed side by side) but becomes more freeform with column selector styling.
Regions are intrinsically freeform but can gain structure through careful
styling. It seems to me that designs that adhere most closely to a standard
newspaper or book layout may be more easily expressed in multicol. But
wilder magazine layouts could require quite a lot of complex column
overrides, and may be easier to comprehend as regions. I think there's room
for both approaches.

One feature that Plan B does not address is the named flow. Being able to
select and redirect content is a powerful building block that I do not think
we have explored fully. I believe that collecting non-contiguous content and
choosing its in-flow location opens up some possibilities that CSS has not
had before. I fully expect that there will be single-region uses of named
flows (but we'll definitely need to add some examples to justify this
claim).

I agree that there needs to be a way to define regions without using
structural elements, and that auto-generation and pagination need to be
addressed. The question is where and when, and whether the current scope of
css3-regions is too minimal. For auto-generation, I think what we have now
is insufficient. Regions should be able to use auto height to display all of
a flow's contents. But that's just the barest minimum - other, more
sophisticated auto-generation mechanisms should follow.

I'm envisioning a road map that includes css3-regions, css3-positioning
css3-pagination, a more comprehensive way of creating pseudo-elements, and a
way of defining and selecting page templates that uses all of CSS to create
complex layouts based on page contents. How do we divide up this work into
manageable modules? Do we have to create one mega-spec that covers
everything, or can we agree on useful smaller steps to take?

Thanks,

Alan

Received on Tuesday, 10 January 2012 05:30:30 UTC