To my mind, whether we should have checksums for other than BASE64
encodings boils down to two questions:
1) Do we ever intend to use Content-transfer-encoding: BINARY?
If BINARY were ever widely implemented, surely it would be desirable to
transfer "binary files" using this encoding. (A "binary file" is any
file which has to make it from sender to recipient without any of its
octets being trashed.) Anything content-transfer-encoding that might be
used to transfer such files through the email system needs a checksum. If
we keep BINARY, we need to define a way to specify such a checksum.
If we are not going to use BINARY, or at least not now, perhaps it should be
removed from RFC XXXX or moved to an "experimental" RFC.
2) Is QUOTED-PRINTABLE an acceptable way of sending "binary files" around?
Certainly it is possible to send a "binary file" around via
quoted-printable, because any bit pattern can be represented.
If a restricted character set is used, QUOTED-PRINTABLE can
be as safe for binary transport as BASE64, and for a very many files,
will result in a more compact (and still somewhat readable) encoding.
It is as easy to define a checksum for QUOTED-PRINTABLE as it is for BASE64.
Keith