Re: [idn] UTF-8 / RACE

[Keith Moore <moore@cs.utk.edu>]

> It is good to hear that the overall WG is keeping in mind that most of the
> world does not understand English well enough to type in that language.  For
> Arabs, we write from right to left, and we have no notion of upper case and
> lower case, plus it is a phonetic language.  This is why a RACE solution in
> the short-term would be not so good for us in the Middle East.  

I don't understand how the last statement follows from the earlier ones.
What does it matter how the IDN is encoded within a protocol, as long as
the application handles the encoding and decoding correctly?

> If the
> members of this WG are seeing the fact of applications being deficient in
> their ability to display RACE-encoded IDN's as a problem, for us in the
> Middle East it is a very big problem.  

I have no doubt that this is true.  But how many of your currently -deployed
applications could display UTF-8 encoded IDNs?  And of those, how many of
them would be able to accept UTF-8 IDNs as input and nameprep them before
doing any queries or name comparisons?

> Arabic and English are very
> different, and our population does not have a strong command of English.  As
> we wait for applications to display RACE properly, it will set us back for a
> long while - that is the crux of the problem I am afraid of.

Is display of IDNs the most important concern, or is it also important that
these applications be able to accept them as input and look them up in DNS?
If the latter two are also important, I don't see why it will take any longer
for these applications to be updated to handle ACE vs. UTF-8.

> For example, the number one Internet application in the Middle East today is
> probably Microsoft Internet Explorer.  It already has support for UTF-8.

perhaps it is able to display UTF-8 text and accept UTF-8 as input, but it 
does not canonicalize UTF-8 names before issuing a DNS query.

> Judging from the emails in this group, Microsoft's claims of "just send
> UTF-8" are not enough.  However, for the entire Middle East, the little
> support that comes with IE 5 for Arabic URL's is a definite positive step,
> and there is momentum there.

Since an ACE is just an encoding of UTF-8, this momentum would apply equally 
to ACE as to UTF-8.  And as you point out, MSIE would have to be updated
in either case.

Keith