Bruce Lilly <blilly(_at_)erols(_dot_)com> wrote:
The amplification factor that Marc Mutz identified could be used as a
DoS attacxk against an MTA as well as against an end user.
Summary: I think the ideal situation would include having
deflate-8bit, deflate-base64, and the escaping/line-breaking method
(separate from compression),
One way to address the both the concern in the first paragraph and
the desire in the second would be to expand the MIME architecture to
incorporate the Content-Encoding field from HTTP.
The Content-Transfer-Encoding would continue to be for the express
purpose of making the message body conform to the transport
requirements. In addition to the quoted-printable and base64
transformations, we would add escaped-8bit (or whatever people decide to
call it).
Compression would be handled separately by Content-Encoding, same as in
HTTP (which already defines gzip and deflate encodings).
The semantics of Content-Encoding are different from
Content-Transfer-Encoding. The latter describes the message body,
not the object it represents. But the Content-Encoding describes the
object itself. With compression handled by Content-Encoding rather
than Content-Transfer-Encoding, MTAs would never need to decompress the
body, because transport agents at most reformat the message, but never
transcode the object itself.
AMC