my comments on draft-ietf-idnabis-bidi-05

["Martin J. Dürst"]

Hello John,

On 2009/09/08 2:22, John C Klensin wrote:
>
>
> --On Monday, September 07, 2009 6:21 PM +0900 "\"Martin J. Dürst\""
> <duerst at it.aoyama.ac.jp> wrote:
>
>> I herewith formally propose to:
>>
>> In 1.4, to make things easier to follow, change
>> "o BN - Boundary Neutral - control characters"
>> to
>> "o BN - Boundary Neutral - control characters (in the context
>> of this document, ZWJ and ZWNJ)"
>
> I'm fairly strongly opposed to this because it means that, if we make
> some other BN character context-dependent in the future, perhaps with
> respect to some newly-added script with other than left-to-right
> behavior, we have to change things in two places. Part of the current
> model, falling out from Unicode-version-independence, is that changes,
> even changes-by-exception, are very isolated. Restricting BN in Bidi to
> ZWJ/ZWNJ would be a fairly significant step and I haven't seen anything
> I'd consider a strong enough argument for doing it. Just my opinion, of
> course.

You are right that if the above can be read as normative, then that 
would be a mistake. So what about:

"o BN - Boundary Neutral - control characters (in the context
  of this document, for example ZWJ and ZWNJ)"


>> In 2., change point 5 from:
>> " 5. In an LTR label, only characters with the BIDI
>> properties L, EN, ES, CS. ET, ON and NSM are allowed."
>> to
>> " 5. In an LTR label, only characters with the BIDI
>> properties L, EN, ES, CS. ET, ON, BN and NSM are
>> allowed."
>
> Maybe, but, if a change is really needed here, my instinct would be to
> say "only characters...ON, NSM, and characters specifically allowed by
> contextual rules, are allowed" or something to that effect.

That might be okay if we go back and verify for every context that it 
doesn't affect bidi behavior. The problem is that there are characters 
other than control characters that are allowed via context, and we have 
to check each of these. Also, it may make implementations somewhat more 
complex; what is expressed simply in terms of bidi categories now 
suddenly becomes a combination between bidi categories and table properties.

Regards,    Martin.

-- 
#-# Martin J. Dürst, Professor, Aoyama Gakuin University
#-# http://www.sw.it.aoyama.ac.jp   mailto:duerst at it.aoyama.ac.jp