At 1:18 AM 12/16/94, Stefan Monnier wrote:
- as Kazuhiko noticed, encryption also requires conversion to
canonical form
I think so, too.
- get rid of the useless "protocol" parameter, or explain to me what good
that parameter will do you, for example, in this case:
Content-type: multipart/signed; protocol=multipart/sig-alternative;
Well, it turns out that I was was of the silent participants in reviewing
earlier versions this spec and pushed back pretty hard about this
parameter. The fact that it is entirely redundant with later information
was quite off-puting (and, in fact, an earlier version of the spec used
different values which would therefore have needed maintenance of
yet-another table.)
The issue has to do with dispatching and the pain of unbounded look-ahead.
Processing by a mime parser is made vastly easier if this key (pun?)
information is replicated at this point.
d/
--------------------
Dave Crocker
Brandenburg Consulting +1 408 246 8253
675 Spruce Dr. fax: +1 408 249 6205
Sunnyvale, CA 94086
dcrocker(_at_)mordor(_dot_)stanford(_dot_)edu