Charles Lindsey <chl(_at_)clw(_dot_)cs(_dot_)man(_dot_)ac(_dot_)uk> writes:
"D. J. Bernstein" <djb(_at_)cr(_dot_)yp(_dot_)to> writes:
Here's a better transition plan:
(1) Teach every message reader to accept UTF-8 in message/rfc822.
Some readers do this already.
(2) Allow message writers to generate UTF-8 in message/rfc822.
It's not a question of "allowing" them to generate such thing. The plain
fact of the matter is that they WILL write such things.
No, that is not right either. They already ARE writing such things, and
they are not going to stop just because the IETF wrings its hands.
The _real_ problem is that they are not using UTF-8.
You just contradicted yourself.
Whether or not clients are inserting random untagged 8-bit character sets
into message/rfc822 messages is another matter entirely. I have seen no
evidence presented that they're currently putting UTF-8 there; in fact,
use of UTF-8 in news messages and I expect e-mail messages as well (which
are not the same thing; maybe you don't understand what message/rfc822 is)
seems quite small.
The reason why they're not is because, as djb points out, there's a
transition that has to happen. *First* you upgrade readers to support
UTF-8, and *then* you allow message writers to generate it. The readers
have, by and large, not yet been upgraded.
--
Russ Allbery (rra(_at_)stanford(_dot_)edu)
<http://www.eyrie.org/~eagle/>