[Mailman-Developers] ASCII Prefixes
Barry A. Warsaw
barry@zope.com
Tue, 23 Jul 2002 13:01:30 -0400
Here's a tricky issue, one with a high rabbit hole probability (which
I'd like to avoid if possible ;).
Say you've got a mailing list that accepts both French and English,
with French as the default language. The default charset for French
is iso-8859-1 and for English us-ascii. Now say you want to add a
prefix to your list, say "[MyList]", and you're users like to send
messages to the list with funny <wink> non-ascii characters in
iso-8859-1.
What Mailman 2.1 cvs will do now is encode the prefix using the
charset for the list's default language. So you'd likely end up with
a Subject: header like
Subject: =?iso-8859-1?q?[MyList]_?= =?iso-8859-1?hello_world?=
or some such. Encoding the prefix can be considered ugly, but
unfortunately we don't really have a way to specify a charset for the
subject prefix (or footers, or other added text) apart from the
default language's charset. So this is /almost/ the best we can do,
even though it's kind of ugly.
OTOH, we could use a simple heuristic to try to keep the prefix
unencoded in the common case: if the prefix contains only ascii
characters, we could choose to leave it in the default us-ascii (or
really the None) encoding. If it contains non-ascii characters, we'd
have no choice but to encode it.
I really don't want to go down the path of providing a gui for
different prefixes for each supported language. Eventually, maybe
that's what we need to do, but I'd like to avoid that for MM2.1. I
think this approach should be good enough, but I'd like to hear
comments, especially from Asian language users.
-Barry