713621 - Inline elements with dir="" should have unicode-bidi: embed

Status: RESOLVED INVALID

(Core :: CSS Parsing and Computation, defect)

Description

[Simon Montagu :smontagu]

Attachment: Patch

This is revealed by rniwa's tests in bug 706194, specifically test_default_bidi_css.html.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

So for [dir="auto"] we already set unicode-bidi: -moz-isolate.  So is this changing behavior only for invalid values of dir, or am I missing something?  (And is there a spec that says how we should handle invalid values?)

Comment 2

[Simon Montagu :smontagu]

(In reply to David Baron [:dbaron] from comment #1)
> So for [dir="auto"] we already set unicode-bidi: -moz-isolate.  So is this
> changing behavior only for invalid values of dir, or am I missing something?

Yes, that's correct.

> (And is there a spec that says how we should handle invalid values?)

http://www.whatwg.org/specs/web-apps/current-work/multipage/rendering.html#bidirectional-text has
 [dir] { unicode-bidi: embed; }
It doesn't seem to be defined other than that.

(By the way, it also has 
 :dir(ltr) { direction: ltr; }
 :dir(rtl) { direction: rtl; }
rather than our
 [dir="ltr"] { direction: ltr; }
 [dir="rtl"] { direction: rtl; }

Is that going to cause any change in behaviour? Will it be worth changing it once we have support for :dir?)

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Since I disagree with the spec here, I posted http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2012-January/034489.html

Comment 4

[Simon Montagu :smontagu]

That got no response, so shall we WONTFIX this and change the tests? I'll attach an extra patch to bug 706194

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Well, it needs a response to get the spec changed... I suppose I should poke Hixie.

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I think http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2012-April/035300.html means this is now invalid.

Comment 7

[Simon Montagu :smontagu]

OK, this is invalid (and the review request on attachment 603424 [details] [diff] [review] is renewed ;-) )