Re: [idn] I-D ACTION:draft-ietf-idn-uri-00.txt

["D. J. Bernstein" <djb@cr.yp.to>]

Keith Moore writes:
> You posted an example where you claimed that your software would have 
> trouble,

False.

> because that software generates unusually long local-parts,
> leaving very little room for domain names.

My mailing-list software often generates local-parts that include the
_subscriber's_ local-part and domain name. The result may exceed length
limits in the _subscriber's_ mail software, or in relays.

> the presumption is that ACE IDNs will cause less trouble than UTF-8
> IDNs.  

Actually, people have been claiming that ACE IDNs will cause no trouble
at all, and suggesting this as a litmus test for IDN solutions.

RFC 2825, for example, says that _some_ of the proposals will cause
failures if software isn't upgraded first. In fact, _all_ of the
proposals will cause failures if software isn't upgraded first.

As for the frequency and severity of failures: What matters is the
software involved in deliveries _where IDNs appear_. People likely to
use IDNs are also people likely to have already moved to 8-bit-clean
software---because 8-bit data shows up all the time in Subject lines,
despite Keith's efforts to force people to encode it.

> But (assuming careful choice of the ACE) few ACE IDNs are likely to be 
> significantly longer than UTF-8 IDNs,

That simply isn't true. A 10-byte limit causes much more trouble for the
ACE proposals than for UTF-8. Try some examples!

---Dan