[Mailman-Users] maiman, problem with greek characters

[Mark Sapiro]

Brad Knowles wrote:

>On 4/27/07, test at social.soc.uoc.gr wrote:
>
>>  I have installed the latest version of mailman in a mail server running
>>  postfix. The problem is with emails that contain greek characters. If the
>>  encoding is iso-8859-7 or windows-1253, all the body text is encoded as
>>  base64. The result of it is that the recipients cannot read the received
>>  emails.
>
>This is not a Mailman problem.  This is caused by either your mail 
>client generating base64, or the message being converted when it hits 
>the MTA.


Actually, it is Mailman (or Python if you prefer). When various
processes (scrubbing attachments and/or adding msg_header and/or
msg_footer) set the message payload with character set iso-8859-7, the
Python email library encodes the payload as base64.


>Get the client or the MTA fixed, and Mailman should handle the result 
>just fine.


Yes! base64 is a standard MIME encoding which the recipient's MUAs
should recognize. Why do the recipients have difficulty reading a
base64 encoded message?


>>  In case a user sends an email with multipart/mixed encoding everything is
>>  ok.


I notice your list post was sent with

Content-Type: text/plain;charset=multipart/mixed

This is completely bogus. There is no such character set as
multipart/mixed. I don't know if your SquirrelMail MUA did this on its
own or you somehow told it to, but it's wrong.

The reason this may work to avoid base64 is that Mailman/Python doesn't
recognize this charset and treats it as unknown resulting in the
messages Content-Transfer-Encoding: being 8bit instead of base64.


<snipped good advice from Brad>


>>  I've added a new line in Defauls.py containing the iso-8859-7 but it does
>>  not work.
>
>Don't edit Defaults.py.  That file will get overwritten in an 
>upgrade.  Edit mm_cfg.py instead.


Absolutely. But, what exactly did you add? Adding arbitrary things that
aren't already defined won't do anything because Mailman won't
reference them.

Adding a new language to the LC_DESCRIPTIONS list via an add_language
line without adding all the translated messages and templates for that
language will just create a mess.

Changing the characterset associated with a language won't affect this
issue, because the character set in this case is the character set of
the message body, not the character set of the list language.

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan