[For this message only, I'm ignoring Ned's issue of whether we need a
signature preservation type at all.]
Blake Ramsdell wrote:
For signature preservation, three proposals exist.
Actually, five proposals exist. Three of them are named
application/mime.
2a. There is the application/mime proposal as it existed three weeks
ago, where the application/mime contents may be transfer-encoded. This
type appears to violate (or amend) the recursive encoding restriction of
MIME.
2b. There is the application/mime proposal which may not be
transfer-encoded. This proposal is better named message/mime-entity.
This type may contain an 8bit or binary object, but is not capable of
tunneling it through a transport which is not capable of handling 8bit
or binary data, respectively.
2c. Same as 3b, but the contained object is prohibited from having 8bit
or binary data at all.
3. application/signed. Less generic, but still interesting to
intermediate processors. Pretty much unreadable by any current mail
agent. MIME tools exist that can process the data.
Like the current incarnation of application/pkcs7-mime, this requires
the MIME parser which processes the resulting data be capable of
handling binary MIME.