Re: [css3-multicol] Review of multicol-block-clip-001

Also sprach "Gérard Talbot":

 > > However, it makes a lot of sense to also make widows/orphans apply in
 > > i multicol layouts -- even in non-paged media, no? Probably, the
 > > multicol spec should address this.
 > orphans : "minimum number of lines in a block container that must be
 > left at the bottom of a page."

That's the CSS 2.1 definition. CSS 2.1 dealt with pages, but not
columns, so it's natural for it not to mention columns. 

Wikipedia, however, mentions columns along with pages:

  "In typesetting, widows and orphans are words or short lines at the
  beginning or end of a paragraph, which are left dangling at the top
  or bottom of a column"

  "Widow: A paragraph-ending line that falls at the beginning of the
  following page/column,"

  "Orphan: A paragraph-opening line that appears by itself at the
  bottom of a page/column."

So, I'd argue that 'orphans' and 'widows' should apply to columns as
well, and (a future version of) the specification should address this.
Therefore, I don't think the tests should punish implementations who
apply this.

 > So, how would (or could or should) this apply to the test in paged media
 > anyway? I am more tempted to remove 'orphans: 1' and 'widows: 1' than to
 > keep those.
 > When I had the vendor-prefix and when in print preview, Chrome
 > 28.0.1500.71 fails the test regardless of the widows and orphans
 > declarations.
 > If the test is supposed to be tested also in page media, then we should
 > create another test and then add the "paged" flag to such test. As is,
 > this test should be passed or failed in screen media. Whether the test
 > passes or fails in paged media really should be in another separate,
 > distinct test.

Removing the explicit "screen" media type seems like an elegant
solution to me. It makes tests simpler, and it means that page-centric
implementations also can run the tests (both Prince and AntennaHouse
have mature multicol implementations).

Requiring a separate test for each media type would turn into a
combinatorial explosion; there are many media types:

Do we really need to make a set of braille test cases and have two
passing implementations in order for mulicol to go to PR?

              Håkon Wium Lie                          CTO °þe®ª        

Received on Thursday, 25 July 2013 07:34:09 UTC