I've come across what looks like an ambiguity in the MIME specs.
Consider the following bodypart within a multipart with boundary "foo":
--foo
Content-transfer-encoding: quoted-printable
=01=02=03
--foo
what is the canonical form of the body part ?
is it: (hex) 01 02 03
or: (hex) 01 02 03 0D 0A ?
(I think it's the former)
in other words, does the fact that the initial CRLF is attached to
the boundary marker and not part of the preceding body part, mean
that the preceding body part (in this case) doesn't end with a CRLF?
Now consider this body part (ending with a soft line break):
--foo
Content-transfer-encoding: quoted-printable
=01=02=03=
--foo
is the canonical form of this body part also (hex) 01 02 03?
or is this just illegal because there's no following CRLF?
Keith