Re: Bidi issues

 > Nevertheless, the example of the glyph sequence "the-arabs=BARA-LA"
 > having two utterly different internal representations remains a strong
 > reason not to allow such local-parts.

But the qlyph sequence

ab.CD.ef (display order) already has two completely different internal
representations:

If it was rendered in an LTR context, the corresponding domain name
(in logical order) is ab.DC.ef

If it's in an RTL context, unless I'm mistaken, the corresponding
domain name is ef.DC.ab

ie in one case the TLD is ef, in the other case it is ab.

As I understand it the way that the IRI draft currently resolves this
is to specify that IRIs are always rendered in an LTR context.

Can I pose another question: if IMA practice were to follow the
currently proposed IRI practice (ie render by simply applying bidi in
an LTR context), then the e-mail address (logical order)
[email protected] would render as:

	[email protected]

Would Hebrew/Arabic e-mail users find this natural?  Or would they
prefer the @-sign to logically separate the address into a localpart
and a domain:

	[email protected]

Further, consider the address [email protected] (logical order).
Part of the rationale for the current design of ToASCII is to allow
constructs such as this.  This would render as:

	[email protected]

Again, is this natural?  If users of RTL languages are confortable
with embeddings such as

	BARA.BEW.com (for WEB.ARAB.com)

then I see no reason why not, but it would be good to get some feedback.

	-roy