Rick,
I'm talking about a
fundamental difference in thinking. Mail, even STMP (that is, on the
wire) isn't a stream of bytes, it's a sequence of records, sometimes
"continued", and we find that one of the rules for continuation is
difficult to implement reliably across all Internet-connected platforms.
I tell you, this would be to everyone's advantage in the long run.
I partially agree with your way of thinking, but not that each *line* is a
record. As this thread has pointed out there is a flaw in using this line of
thought, that being the equivalence of CR LF SPACE CR LF and CR LF CR LF after
converting from RFC822 to your record-oriented format.
I think that a better approach for new software would be to treat the *entire
header* as a single record and the *entire body* as another record. Keep the
CRLF's in there and reduce the overhead of converting back and forth. It also
reduces the dependence on which O/S you are developing on. And while you're
at it, you could break up multipart-MIME messages into several records (at the
boundary markers obviously) for easy manipulation and recursive translation
to/from 7bit....
Michael D'Errico
Software.com, Inc.