I wonder if the right fix here would be to allow binary CTE of
message/rfc822 content. MH originally allowed only 7bit encoding of
multipart and message content types. By the time nmh was put under
cvs control, 8bit and binary had been added for multipert. Commit
d057cb added allowance of 8bit for message content type. Should we
add binary?
Actually ... I think you're right that we should add binary. The concern
about q-p or base64 was double encoding, but that's not a concern for
binary.
RFC 2045 § 6.4 says "Even more severe restrictions [than 7bit, 8bit,
or binary] apply to some subtypes of the 'message' type.", but I don't
see where those more severe restrictions are listed.
Looks like that's RFC 2046, § 5.2. For message/rfc822, binary is
explicitly permitted. But for message/partial, 8bit and binary
are prohibited.
--Ken