Re: Bidi issues

Ok, here's what I've got so far. It's very rough and ready, but it may
be of some interest.

http://www.gnomon.org.uk/bidi-ambiguities.txt

Not sure how useful it is, and it's definitely unfinished (eg the
analysis or IRIs and IMAs promised in the introduction doesn't exist
yet).

I don't think any of this is really of direct relevance to the IDNA or
IMAA protocol specifications, though it could eventually form the
basis for a set of recommendations that could go into a BCP...

BTW, rereading my previous post, I see I was somewhat vague as to what
I was suggesting:

I was proposing that there be no bidi restrictions at all on the
localpart (ie use a variant of nameprep without the stringprep bidi
restrictions).

I'm currently tending towards the opinion that the benefit of bidi
restrictions doesn't justifies their cost in terms of disallowing
useful IMAs, particularly given that no set of bidi restrictions I can
currently envisage is going to remove the need for administrative
procedures to avoid allocating potentially ambiguous names.

I'd welcome feedback.

Usual disclaimer, I'm no bidi expert, etc...

	-roy