[Mailman-Developers] Re: mangling of subject header by outlook or mailman

Tokio Kikuchi tkikuchi@is.kochi-u.ac.jp
Mon, 12 Nov 2001 22:19:16 +0900


totschnig.michael@uqam.ca wrote:

> 3) a user of outlookexpress 5 replies both directly and via the
> list. the direct reply has the following header:
> Subject: =?iso-8859-1?Q?Re:_=5BTest=5D_test_de_deboguage_des_=E9_et_=E8_dans_le_he?=
>         =?iso-8859-1?Q?ader_subject?=
> which still gets correctly decoded by mail readers.

How other mailers do with this phase of subject ?

This problem has long been known to Japanese mailing list owners
and we solved this problem by decoding the MIME header _before_
adding the list prefix and then re-encoding the subject into MIME.

I wrote a module to convert the incoming Japanese mail into EUC
while decoding the MIME headers. All the outgoing messages should
be encoded in MIME scheme and I also wrote one to do this.
They are integrated into Japanese Mailman;

May be we should `canonicalize' all the message upon entering 
the Mailman processing pipeline and encode into appropriate charset
when going out through SMTP/NNTP/HTTP.

How about canonicalizing into Unicode, since Japanese and Chinese
must be properly treated ?