[Gen-art] LC review: draft-ietf-idnabis-bidi-06.txt
matial at il.ibm.com
Mon Oct 5 22:49:46 CEST 2009
I concur with what Erik van der Poel wrote.
I also agree that practically there is currently no difference whether CS
characters are allowed or not by the bidi rules (since they are disallowed
by the Table anyway) and that there is very little likeliness that there
might be a difference in the future (in the theoretical case that a new CS
character would be added to Unicode which would be allowed in IDNA
I suggest to add a note in section 1.4 "Terminology" in the description of
the CS Bidi property saying that currently none of these characters is
allowed in IDNA labels (analogous to what is said about LRE, LRO, RLE,
RLO, PDF). That would make the mention of CS in bidi rules #2 and #5
Shalom (Regards), Mati
Globalization Center Of Competency - Bidirectional Scripts
Phone: +972 2 5888802 Fax: +972 2 5870333 Mobile: +972 52
Erik van der Poel <erikv at google.com>
Andrew Sullivan <ajs at shinkuro.com>
idna-update at alvestrand.no
Re: [Gen-art] LC review: draft-ietf-idnabis-bidi-06.txt
idna-update-bounces at alvestrand.no
Yes, this change occurred between bidi draft versions 3 and 4, and is
due to Mati's proposal. I believe he came up with that suggestion
based purely on the bidi criteria (grouping, uniqueness), and not
based on what was allowed/disallowed in the Table draft.
I think it's OK to leave CS in there, even if the Table spec excludes
all of them in the current version of Unicode. Future versions of
Unicode might introduce new CS characters that we want to allow in
IDNs, though that seems unlikely.
On Mon, Oct 5, 2009 at 12:40 PM, Andrew Sullivan <ajs at shinkuro.com> wrote:
> On Mon, Oct 05, 2009 at 12:35:29PM -0700, Kenneth Whistler wrote:
>> > At the moment, I can't reconstruct what we were thnking when we
>> > to allow it.
>> Neither can I.
>> > The parenthetical remark can go without great loss; nothing depends
>> > it. But I wondre why we are allowing CS at all.
>> I think the text of idnabis-bidi-06.txt should just remove CS
>> from the list of characters allowed in either RTL labels or LTR labels.
> I seem to recall a thread some time ago (I didn't rummage through the
> archives to find it) that insisted that bidi not exclude any category
> that was not _required_ to be excluded by the bidi rule. The idea was
> that a given category should only be ruled out by bidi if it is
> necessary for the bidi rule, and for no other reason. Any other
> exclusions were to be covered by tables (and only tables).
> This was all part of the argument for bidi standing on its own.
> Again, I'm going exclusively from memory here. But if I had to
> construct an argument for leaving it alone, the above would also be
> the line I'd take. (That said, I have no strong opinion one way or
> the other.)
> Andrew Sullivan
> ajs at shinkuro.com
> Shinkuro, Inc.
> Idna-update mailing list
> Idna-update at alvestrand.no
Idna-update mailing list
Idna-update at alvestrand.no
More information about the Idna-update