Comments on IDNA Bidi

[Kenneth Whistler]

> > Yes, but as I indicated, all the explicit embedding stuff is
> > irrelevant. All those codes are categorically ruled out by
> > RFC 3454, and I don't think anything we are proposing allows
> > them back in.
> >   
> It's only irrelevant if we can ban LRE and RLE *around* the domain
> names, as well as within them.

For large enough scopes of "around", of course you can't ban
them. But I think IDNA needs only be concerned about the
labels themselves and their immediate syntactic contexts in
concatenations of labels for URL's and the like.

> What I get from this statement is that the paragraph
> 
> <sor>this is a <rle>ABC<pdf> domain.NAME<eor>
> 
> is a case that we don't need to test for. People who do that will get
> weird results, but it's not a problem.
> 
> Is that what you are saying?

Yes.

> 
> It makes life a LOT simpler if this is OK. Not quite as simple as
> IDNA2003 assumed, but a lot simpler.

Well, I'd like a confirmation from Michel and Mark, who
actually have bidi implementations to hand and deal with
URL's and IRI's actively in application contexts, but
I do think that IDNA needs only to deal with a rather
constrained set of bidi issues, and not with the total
universe of unconstrained embedding of text in totally
generic contexts.

--Ken