Neil Katin writes:
I checked with our Japanese localization people. They said that
JUNET doesn't prohibit japanese in the headers;
I just had a quick look at the JUNET guide (which also contains the
JUNET rules), and I couldn't find anything that prohibits Japanese in
headers. However, there is a guideline that says that you should use
headers of the form
Subject: next meeting (In Kanji)
to inform people that Japanese is included in the message. (Of course,
this doesn't prove that Japanese cannot be used in headers, but...)
I know for a fact
that our (Sun's) current implementation supports this feature,
The keyword here is "current". This means that older systems don't
support this feature. Remember that not everyone has the newest
system.
and that customers are using it
Perhaps your customers are using it within their own organization,
much like the use of 8-bit SMTP, which doesn't interoperate with the
rest of the world. (Why don't you try sending a message with escape
sequences in the header to THIS mailing list. :-)
Although I sympathize with the people who want to use character sets
other than ASCII in headers, I cannot see a clean solution to this
problem. Perhaps we need to compromise. How about:
Header-Mnemonic-Intro: 38 (&)
Header-Substitution: 1=text/iso2022=base64=f2lMeRs2k9JjlmN
From: Keld J&o/rn Simonsen <keld(_at_)dkuug(_dot_)dk>
To: Taro Suzuki *1 <suzuki(_at_)foo(_dot_)co(_dot_)jp>
Subject: new locale definition
Here we use Mnemonic for the kind of text that it is best suited to,
i.e. alphabetic languages, and we allow romanized Japanese (Taro
Suzuki) for non-Japanese readers and Japanese users of old mailers,
together with a substitution method for new mailers (i.e. the "*1"
corresponds to the "1=..." stuff). We put the "Header-" headers above
the "From:", etc old-style headers, so that old mailers that skip
headers other than To:, From:, Subject: and whatever, can skip the
ugly Header-* headers when displaying.
Yeah, it's disgusting, I know...
Cheers,
EvdP