Re: [CSS21] Overconstraint issue in Section 10.3.8

On Wednesday 2005-01-12 15:09 -0800, Grant, Melinda wrote:
> When an object is absolutely positioned using the 'bottom' and 'right'
> properties and direction is ltr, step 2 of the current algorithm in
> Section 10.3.8 seems to inappropriately set 'left' to the static
> position when the intrinsic width of the object implies a different
> left position.  (Essentially the 'right' offset is overridden by
> setting 'left' to the static position.)  And the analogous issue for
> rtl also applies, I believe.

This seems like a reasonable complaint.  What do existing
implementations do?

The alternative wording I can think of that would fix this problem and
seems most likely to match implementations if they aren't doing what the
spec currently says is replacing (2) and (3) with:

If both 'left' and 'right' are 'auto', then:
  * If 'direction' is 'ltr', replace 'left' with the <a>static position</a>.
  * If 'direction' is 'rtl', replace 'right' with the <a>static position</a>.

10.3.7 and 10.6.5 have the same problem.  I think 10.6.4 does not (which
might suggest another alternative).

(It's also a possibility that the group agreed to changes to these rules
that never made it back into the spec.)

-David

-- 
L. David Baron                                <URL: http://dbaron.org/ >