It's recently come to my attention that a number of user
agents are broken, such that if you send them a message
that contains one or more RFC 2047 encoded-words, and
the recipient generates a reply, the reply contains the
*decoded* form of the encoded-word....
thus using non-ascii characters in the message header.
This certainly violates several standards, is subject
to ambiguous interpretation, and appears to cause a
variety of operational problems.
A related problem is with UAs that accept non-ascii characters
in address books (for the name of the recipient), and don't
encode them in RFC 2047 when sending a message.
I'm wondering how widespread this problem is, and what, if anything,
can be done to encourage implementors to implement RFC 2047 more
carefully.
Keith