On 5-feb-04, at 22:19, Keith Moore wrote:
Suppose we're both at an IETF meeting or such, and you want to email
me
the latest list of goals, which is now 25 MB. If your setup is
anything
like mine, the message will first travel back to a place close to your
office, then to mine and finally back to whereever the meeting is
held.
This is enough of an inconvenience to create some special case logic
that makes it possible for the file, if not the rest of the message,
to
be delivered from your computer to mine directly.
Because we as users experience the transfer as too slow or unreliable?
Slower, less reliable, uses unnecessary bandwidth.
I think it would be useful to sharpen our thinking to work (for at
least a while) under the assumption that with mailng everything that
is
done with SMTP, POP, IMAP and HTTP is now done using a single
protocol,
so there is no artificial dichotomoy between servers and clients.
a) I don't know how to translate that into a user goal
It doesn't.
b) actually I'm inclined to think we need more specialization rather
than less specialization.
That could very well be. But we should base this decision on
engineering tradeoffs and not limit ourselves to replacing one of the
protocols that play a part in email at a time.