Erland Sommarskog wrote:
Punycode is useless, because it is not compatible with anything. Sure,
nothing breaks, but a newsgroup name must be display well and a name
in Punycode gets wrecked beyond recognition, even for Latin scripts.
Unless, of course, you have a reader that understands Punycode, but
there is no such thing today. So it is not very likely that any
hierarchy would actually use it.
Our disagreement comes down to the definition we're each using for
"compatible". I would say that punycode is compatible with all MIME,
DNS, and Internet application protocols, because it is 7 bits, and so it
transits them unchanged. (In fact, it only uses LDH - letters, digit,
hyphens.) Now, it's true that the non-LDH characters encoded in
punycode look like gobbledygook on non-punycode-aware readers. But, the
punycode text is not going to be corrupted by gateways or non-punycode
aware user agents. User agents that understand it can provide good
i18n; the rest of the agents still work fine.
And though no news reader currently supports punycode, it's difficult to
imagine that it will not be ubiquitously deployed within a year or two,
as it will be necessary to display international domain names (IDNs).
IDNs will be supported widely before usefor becomes a proposed standard,
especially at usefor's current rate of progress.
By contrast, raw UTF-8 may work fine for some newsreaders (like
Mozilla), but it is *incompatible* for others, and especially for
gateways (like IMAP servers). Incompatible means that it breaks
software that had been correctly designed to work with previous versions
of the standard.
Note that, as in this case, elegance and compatibility are often at
odds.
- dan
--
Dan Kohn <mailto:dan(_at_)dankohn(_dot_)com>
<http://www.dankohn.com/> <tel:+1-650-327-2600>