Re: Mail Path MTU discovery?

On Fri, Feb 13, 2004 at 08:50:01PM +0100, Hadmut Danisch wrote:
> Or should we have a hard size limit? When you transmit 
> 4 GB, and the last byte fails, you need to transmit it again.

There is absolutly no reason why this has to be the case. HTTP can
resume failed download, SMTP can't, POP can't, but the future mail-ng
protocol can be designed to allow this.

I see several reasons for limiting message size today:
- No resume mechanism for failed transport.
- Limits on spool size.
- Limits in mailbox size. 
- Also any mail system knows that it might have to forward the mail on to a
  different mail system which might impose a limit, so the first mail
  system might get stuck with a large message. So it imposes a limit of
  its own to reduce the chance of this.

Those reasons are all different and we can think about how to solve each
of them.

For instance the "spool size" limit might be obsolete if we can deliver
certain mails directly into the mailbox without them ever beeing in the
spool. I know of no mail system who can do this today, but that doesn't
mean that it's not possible.

Jochen
-- 
Jochen Topf  [email protected]  http://www.remote.org/jochen/  +49-721-388298