Re: RFC 1426 problem and solution

[John C Klensin <KLENSIN@xxxxxxxxxxxxxxx>]

Just to add one or two things to the comments that have already been made...

In addition to the issues associated with getting rid of long lines, there
is the little matter of how an SMTP server can tell that a message has
ended.  The current rule involves the sequence CR LF . CR LF, and that rule
(and the period-stuffing required of the sender) is pretty easy to handle
in a line-oriented context.  If you don't have "lines", then you need a
more complex set of detectors and rules about how to find end of data.

Of course, for "true binary" the right answer is to do it by count, rather
than by parsing.  But that really is another piece of protocol, and
not "8BITMIME", as others have pointed out.

There are also a number of questions about why you would want to transfer
pure binary over a mail connection and whether, if you did, you would 
really want a mail-header structure (inherently line-oriented text) 
wrapped around it.

It is all a lot more of a mess than it appears on first (or third) glance.

     john