> if a message has the content-type field mentioned in the subject,
> which seems valid as far as I can tell, then which setting wins?
> Should a compliant mail reader enable or disable magic while reading
> such a message?
It's arbitrary given that parameter order is insignificant.
two gut reactions:
1. if a content-type wants to define some behavior in which it makes
sense to use the same parameter more than once, I don't recall anything
in the MIME spec that prohibits that, and I can think of a few cases
where it would be useful. e.g.
content-type: message/external-body; access-type=url;
could specify multiple locations at which the desired content could be
It may not be explicitly stated that duplicate parameters aren't allowed, but
given that parameter order is required not to be significant (RFC 2046 section
1) is is hard to come up with any other result that makes sense.
This issue came up when the parameter encoding extensions were being developed
and I added some text specifying that parameters are to appear only once to my
working copy of the MIME specification at that time. Unfortunately the
specification hasn't been published since.
however, I suspect that this would break assumptions made by some
implementations, and for that reason I'd probably object to a proposal
for a content-type that tried to use the same parameter more than once.
2. there are so many ways to do meaningless or stupid things in email
that I doubt it's productive to try to specify the "right" behavior in
every case. only the most common errors seem worth the trouble.
Absolutely. The right thing to do here is to make it clear parameters
can only appear once when MIME is next revised. Hopefully I can get that
particular ball rolling sometime soon...