Re: bidi-004 probably invalid

On 10/13/2010 09:24 PM, L. David Baron wrote:
> I think the following tests:
> http://test.csswg.org/suites/css2.1/20101001/xhtml1/bidi-004.xht
> http://test.csswg.org/suites/css2.1/20101001/html4/bidi-004.htm
> are invalid because of the white space following the "kkk".  Based
> on the white-space processing rules in section 16.6.1, I believe
> that the sequence: "kkk ‬ mmm" should have the first space
> honored and the second space ignored.  (I don't think this behavior
> is ideal for authors, but it is what I think the spec says, and what
> we implement.  We should perhaps consider changing it.  I should
> probably email www-style once I find the other tests that have the
> same issue; I think there are a bunch of others.)
>
> I think the test would be valid if that single space (between the
> "kkk" and the PDF) were removed.

I'm not sure I actually agree here.

http://www.w3.org/TR/CSS21/text.html#white-space-model

Treating multiple spaces as a single space for simplicity, on the
last line we have

_lll_kkk_[PDF]_mmm_nnn_ooo_    <- text
RRRRRRRRR     NLLLNLLLNLLLN    <- directionality

White space collapsing removes the space after the PDF

_lll_kkk_[PDF]mmm_nnn_ooo_

(This is 16.6.1 part 1. Now we move to part 2.)

Then bidi reordering takes effect, with a left base directionality,
and we lay out the line with the following reordered contents:

_kkk_lll_[PDF]mmm_nnn_ooo_

As we lay out the line we remove the spaces at the start and
end of the line, resulting in

kkk_lll_[PDF]mmm_nnn_ooo

which is what the control rendering suggests.

So I think the test is correct as-is.

~fantasai

Received on Sunday, 5 December 2010 09:35:40 UTC