D. J. Bernstein wrote:
Bruce Lilly writes:
For gateways to work at all, either the user agent must preserve charset
and language information in some way that it can be used by gateways
[ or it must use RFC 2047 ]
Specifying ``UTF-8 everywhere'' removes the need to add charset tags to
individual pieces of data. The message writer converts the local charset
to UTF-8 and then discards the local charset information.
That would only work if it could be absolutely guaranteed that no untagged
data in any other character set might appear. No such guarantee is possible,
and in fact Usenet abounds with untagged charsets, of which, according to
Andrew Gierth, "no significant amount" is utf-8.
As for languages: Exactly which gateways use language information?
Anyway, the notion that a gateway _has to_ use language information---
information that almost no messages have---is patently absurd.
Who said anything about gateways having to use language information? I
certainly did not. But the receiving UA may make use of such information
(e.g. in a Subject field), and if the language information is discarded by
a generating UA, no gateway can hope to reconstruct that information unless
it is conveyed to the gateway by the generating UA by some mechanism. Hence
the requirements excerpted at the top of this message.