From the current draft, it appears that a hard line break is defined to be
equivalent to a CRLF sequence. A hard line break is similarly encoded in
base64. Thus there is no inconsistency in the current draft. This also
aligns quoted-printable and base64 with the three other encodings, namely
7bit, 8bit, and binary.
When we had a record separator defined in base64 and quoted-printable we
had a major inconsistency between these two encodings and the rest. In
particular the notion of a out-of-band record separator is difficult to
define in the binary encoding. And, as I said before, the handling of
integrity checks was needlessly complicated. This becomes apparent when
you actually try to implement MIME.
If you find this analysis condescending, well, that's too bad. It is not
intended to be.
If you have a problem with the wording in the current draft, I suggest you
provide an alternative wording that makes it clearer.
Ned