Re: MIME vs. DIME vs. ...

>It's not particularly difficult to create separators that
> are guaranteed not to appear in the message body, simply use
> a cryptographically secure PRNG to generate a suitably
> large random number (160-bits should be plenty).

The probability that a well-chosen random generator will
produce a string that is in the body is very low, I agree.

But I was thinking of a different metric: the probability
that a typical software engineer would use an incorrect
algorithm. I think that probability is high -- maybe 50%.
(Or at least, I've seen a lot of stupid MIME boundary
generators.)

Similarly, while fast string searching algorithms are
well-known, they're hard to get right.

So here's a use case for DIME: programmers
are more likely to get it right!

I still like the herriott draft as having many of the
'simplicity' qualities of DIME but being more explicitly
an equivalent concrete structure for MIME multipart.

And there are many use cases for interleaving.

Larry
--
http://larry.masinter.net