Re: Proposal for isolation characters in Unicode and the unicode-bidi:isolate and unicode-bidi:plaintext definitions

I have compared the content of
http://www.w3.org/TR/2012/WD-css3-writing-modes-20120501/ with that of
http://dev.w3.org/csswg/css3-writing-modes/, and I do not think that the
changes are at all related to this thread. The isolates are still described
in the terms of a higher-level protocol that treats the isolate's content
as if it were in a separate document. The list of major changes does not
make any reference to Unicode isolates either.

Please note that using the new codepoints instead of treating the isolate's
content as if it were in a separate document does introduce two significant
changes in behavior:
- Paragraph breaks (e.g. <br>) within the isolate break the paragraph
within which the isolate appears. This appears to be an improvement over
the spec as it stands.
- The embedding level goes up inside an isolate, instead of being reset to
0 or 1. This is not an improvement over the spec as it stands, but is
unavoidable if isolates are implemented in terms of Unicode codepoints.

I am not pushing to change the CSS spec now, before the new isolate
codepoints have been approved and added to Unicode. I am not at all sure
that it can change in this respect until then.

However, I do want to make sure of two things:
1. That the CSS spec changes I outlined in my original message look
reasonable, so that if the new Unicode codepoints were to be approved
today, the spec could be changed as outlined.
2. That the CSS spec of unicode-bidi:isolate, isolate-override, and
plaintext will not become frozen before it has been changed to use the new
Unicode codepoints.

The great benefit for changing the spec to be based on the new codepoints
is that it will make it much easier to implement isolates. Current browser
implementations have different, serious, difficult-to-solve bugs stemming
precisely from the fact that they had to work around the UBA instead of
just letting it do its job.

It would also be pretty terrible if the Unicode and CSS specs offered very
similar features that nevertheless behaved quite differently in certain
perfectly valid cases.

I would be very happy to discuss the details of the CSS spec changes I
proposed. (Actually, if anyone is interested in such a discussion, I should
probably start by update that proposal given that both the CSS spec and the
Unicode isolates proposal have changed in the meantime.)

Aharon


On Tue, Jun 26, 2012 at 9:01 AM, fantasai <fantasai.lists@inkedblade.net>wrote:

> On 05/17/2012 05:16 AM, Matitiahu Allouche wrote:
>
>> I am in favor of Aharon Lanin's proposal for 3 new characters: LRI, RLI
>> and FSI, with Martin Duerst's addition of a fourth
>> character to terminate the scope of the last unclosed RI, RLI or FSI.
>>
>> I also agree that the HTML/CSS behavior for the BDI element should be as
>> similar as possible to the behavior of those 4
>> characters in plain text.
>>
>
> I've attempted to work this proposal into the Writing Modes specification.
> Here's the editor's draft:
>   http://dev.w3.org/csswg/css3-**writing-modes/#bidi<http://dev.w3.org/csswg/css3-writing-modes/#bidi>
>
> ~fantasai
>