[Mailman-i18n] charset impovements

[Martin v.Löwis]

barry@python.org (Barry A. Warsaw) writes:

> So the b5 changes look good?

I haven't actually looked at the changes yet. 

I just tested them, and there are still problems. In particular, the
generation of multipart messages seems to corrupt data.

In my recent Test1 message,

http://mail.python.org/pipermail/playground/2002-November/000028.html

the additional header was transmitted as

---------------------------------------------
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

Cyrillic header text: ÆÕÎËÔÉÏÎ
--------------------------------------------

I think the MIME-Version does not belong there (not sure though),
and the charset is wrong (should have been koi8-r).

Test message 2 was worse:

http://mail.python.org/pipermail/playground/2002-November/000029.html

should read

'This message should be encoded in koi8-r.\r\n\xc1\xc2\xd7\xc7\xc4\xc5\xd6\r\n\r\n'

instead. As you can see in the .mbox file, the original message still
had the encoding as base64 and koi8-r. The archiving incorrectly
displays the base64 data; the message sent out misses the charset
on the primary part (even though the transfer encoding was included).

Test message 3 was processed correctly, MIME-wise, except that header
and footer have the encoding incorrect.

Test 4 is both archived incorrectly and sent out incorrectly.

So I do think there are considerable improvements, just new bugs as
well.

Regards,
Martin