[Mailman-Users] Re: Mime in subject ... more ÄÅÖäåö

[Jeremy Hylton]

Barry asked me to look at the small hack for pipermail.py that
improves decoding support for message headers.  The proposed change is
in the wrong place; a change that effects that way the HTML output is
generated should be in HyperArchy.py.  (This isn't obvious.)

HyperArchy.py already processes messages with encoded words in the
subject and decodes them when possible.  There may be a bug in the
logic there, which I would need some help to debug.  The decoding
problem is straightforward, except that a mail archive may contain
messages encoding using many different character sets.  The message
index can only produce HTML that uses a single character set, so it
has to pick one and go with it.  If it picks ISO-2202-JP, then it
should not decode ISO-8859-1.  It tries to pick the most frequently
occuring charset.

If you are experiencing a problem, my first guess would be that
HyperArch has concluded that a particular message's charset does not
match the one used for the index page.  

If there is a specific problem, can someone point me at a mail archive
that appears to be broken?  If you are looking at a particular message
that contains endecoded subject lines, can you send me the HTML page?
The key question I would have is whether the charset in the <META
http-equiv"Content-Type" ...> tag matches the charset in the subject
line.

Jeremy