W3C home > Mailing lists > Public > www-style@w3.org > June 2010

Re: [CSS21] bidi, text-align, and list markers

From: Peter Moulder <peter.moulder@monash.edu>
Date: Thu, 03 Jun 2010 17:35:40 +1000
To: W3C style mailing list <www-style@w3.org>
Message-id: <20100603073540.GA3320@bowman.infotech.monash.edu.au>
On Wed, Jun 02, 2010 at 04:28:07PM -0700, Tab Atkins Jr. wrote:

> On Wed, Jun 2, 2010 at 11:25 AM, Ambrose LI <ambrose.li@gmail.com> wrote:
> > On 2 June 2010 14:12, Alan Gresley <alan@css-class.com> wrote:

> > But wouldn't it be very counterintuitive to have <li dir=rtl> mean
> > something different than <li><span dir=rtl> ?
> >
> > As an uninformed author (i.e., one that is not following this list
> > closely), I find it very surprising that I'll have to use a child
> > element when I can set the direction in the list item.
> 
> I find it relatively clear.  The ::marker is a child of the <li>, so
> @dir on the <li> affects it, but @dir on a child of the <li> doesn't.

(Incidentally, CSS2.1 has no mention of ::marker, and I'm not aware of text in
 CSS2.1 indicating that the marker is a child of <li> when 'list-style-position'
 has its initial value of 'outside'.  However, CSS2.1 does make clear that the
 <li> generates the marker box, which gives much the same argument as to
 surprisingness.)

While I agree it's understandable why it renders that way once you've noticed
it rendering that way, I also agree that an author may well not expect that
consequence of having unequal 'direction' values among list-items of what the
author considers to be one list.

Some other relevant points:

  - What the author considers to be one list may be affected by counter scope
    rather than containing block.  It may be rare that this makes a difference,
    but it does mean that the proposed change only makes a rare problem problem
    rarer rather than eliminating the problem.

  - Having marker position be determined by the list-item's 'direction' rather
    than "the list"'s 'direction' is at least consistent with list-style-type
    behaviour.  (Though I'm not saying that this would make it unreasonable for
    these two to be inconsistent.)

pjrm.
Received on Thursday, 3 June 2010 07:36:11 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:28 GMT