There will never be. The people of East Asia are not going to adopt an
arbitrary system conjured by a bunch of ignorant white folk. A crude way of
putting it, but true.
So if a message arrives from a Russian professor of Japanese to a Japanese
professor of Russian, and the Subject and body of the message are in a
mixture of Cyrillic and Kanji then the Subject will be OK. Using
Encoded-variable scheme [I presume you are an Encoded-variable supporter]
the Kanji can still be in ISO-2022. However the message will be rejected
by the Japanese computer with "Your message can not be accepted here because
the body is in a universal character set and we can't grok that because it
was not invented here". Actually I feel the Japanese are the last people
in the world to take a silly NIH attitude.
Now I suspect that there is a misunderstanding. The universal character set
header proposal doesn't require a universal character set to be used
everywhere. There is nothing to stop the Japanese using
Header-charset: iso-2022 / 7bit
internally. They can even send messages outside like that and the only
disadvantage of doing so is that if some other e-mail programs elsewhere
want to combine headers from this message with headers in Cyrillic from
some other message then they are going to have to know how to upgrade
ISO-2022 to the universal character set.
I think that it will be a lot easier for existing Japanese software that
puts iso-2022 in headers to be changed to just add a "Header-charset:"
header. The alternative is to generate a variable for each bit of iso-2022
and create a heap of Encoded-variable headers (and then the other end has
to undo all that). And what are the encoded variables names generated by
Nathaniel's algorithm going to look like? [Not that that matters in theory].
Bob Smart
P.S. I put a message on the network news trying to find out how prevalent
it is for software to have problems with quoted phrases with funny
characters in them. I got a reply describing problems with quoted local-parts.
This re-enforces my suspicion that people are remembering problems with
quoted local parts when they say there are problems with quoted strings.
I think we should have evidence of software in use today that is broken by
quoted phrases before we even consider taking such potential problems
into account. It is bad enough going out of our way to allow for broken
software without going out of our way to allow for illusory broken software.
So if anybody has any examples please let me (or us) know.