[Mailman-Users] Attachments are unexpectedly re-created.

[Masaharu Kawada]

Hello Kikuchi-san,

As far as I checked the error messages and compared with each one,
The problem of my customer's seems to be almost the same as one
posted to the thread written in the last Kikuchi-san's email.

There is one thing that I wonder, is that if the charset.py is done with
that patch, what the current digest.mbox exsisted under lists/<listname>
are supposed to be? Do they need to be deleted or not? My point on
this is that once the patch is done, is there nothing else to do to fix this
problem?

Thanks in advance,

Best Regards,

Masaharu Kawada wrote:
> Hello kikuchi-san,
>
> Thank you for your response with a precious information.
> Since I am a beginner for mailman, any comments/suggestions
> such as this kind of thing are very helpful for me. Including
> what Mark-san have been doing for me and you have done
> for me as well, I really appreciate the cooperation from  this list.
>
> As for the problem on my question, I will look into the infomation
> that you just gave me, and later on I will compare my customer's
> with one posted in that thread you provided.
>
> Thanks a lot!
>
> Best Regards,
>
> Tokio Kikuchi wrote:
>   
>> (10/03/03 16:54), Masaharu Kawada wrote:
>>   
>>     
>>> Hi Mark-san,
>>>
>>> I got a reply from the customer, but it was not possbile for the customer
>>> to provide the whole digest.mbox to us due to their policy.  However,
>>> sevral parts of the error log and digest.mbox's content have been provided.
>>> As for the error messages, please see the attachment.  And as for the
>>> digest.mbox which might be the problematic one, I see lots of messages look
>>> like below.
>>>
>>>     
>>>       
>> Hi, Kawada san,
>>
>> The problem is that the Japanese mail users/MUA developers use CP-932
>> charset as Shift-JIS and its derivatives (ISO-2022-JP/EUC-JP).  CP-932
>> contains more (extended) characters than Shift-JIS while Python codec is
>> strict on the latter.  The characters like circled numbers fail to be
>> decoded in unicode and cause error.
>>
>> Similar errors are reported in Japanese Mailman users, like the thread
>> starting from:
>> http://mm.tkikuchi.net/pipermail/mmjp-users/2009-February/002487.html
>>
>> A workaround is to patch charset.py in Python email library as:
>>
>> --- Lib/email/charset.py	2009-09-22 08:59:56.000000000 +0900
>> +++ Lib/email/charset.py.orig	2009-09-22 08:58:36.000000000 +0900
>> @@ -264,8 +264,7 @@
>>      def convert(self, s):
>>          """Convert a string from the input_codec to the output_codec."""
>>          if self.input_codec != self.output_codec:
>> -            return unicode(s, self.input_codec, 'replace'
>> -                     ).encode(self.output_codec, 'replace')
>> +            return unicode(s, self.input_codec).encode(self.output_codec)
>>          else:
>>              return s
>>
>>
>>
>>   
>>     
>
>
>   


-- 
-------------------
Masaharu Kawada
Associate Technical Support Engineer
Red Hat K K
Ebisu Neonato 5F
1-18 Ebisu 4-chome, Shibuya-ku
Tokyo 150-0013, Japan
Direct: +81-3-5798-8482