header encoding problem, multi-lingual text

[Katsumi Yamaoka <yamaoka@jpl.org>]

Hi,

Say I reply to a Chinese friend, who reads and writes Japanese as
well, with a subject like this:

Subject: Re: 答复: No Gnus T-シャツ

"答复:" seems to mean "Re:", so I remove the first one:

Subject: 答复: No Gnus T-シャツ

I realized today that this subject of a message sent to the recipient
is not decoded fully if the recipient uses M$ Outlook; it is shown
in that mailer as follows:

Subject: 答复: No Gnus T-･$B%7%c･(B･$B%D･(B

What happens is that Gnus encodes the subject into[1]

Subject: =?utf-8?B?562U5aSNOg==?= No Gnus =?iso-2022-jp?B?VC0bJEIlNyVjGyhC?=
 =?iso-2022-jp?B?GyRCJUQbKEI=?=

and M$ Outlook decodes it from a charset that is first found in
those encoded words, I believe.  This must not be Gnus' fault,
however it gets to be a real annoyance; that broken subject will
never be fixed while repeating of replies each other.

Then, I made a change in rfc2047.el's header encoder[2] so that
it unifies the charsets into a single one used to encode the
whole text in a header as much as possible.  As for the subject,
containing Chinese and Japanese texts, now `utf-8' will be chosen.

[1] In the Japanese locale, `mm-coding-system-priorities' defaults
    to `(iso-8859-1 iso-2022-jp iso-2022-jp-2 shift_jis utf-8)',
    so `iso-2022-jp' is applied to Japanese text and `utf-8' is
    applied to Chinese text.

[2] `rfc2047-encode-message-header'.  That is an aliasee of
    `mail-encode-encoded-word-buffer'.