On Tue, 2005-05-10 at 23:34 -0400, Sam Michaels wrote:
> On 5/6/05, Philip Hazel <ph10@???> wrote:
> > Exim implements "delay = 5m" in the most trivial possible way. It waits
> > for 5 minutes. :-) I am not sure how easy it is to code "wait for 5
> > minutes or until the connection goes away". It is straightforward to
> > code the opposite, i.e. "wait for 5 minutes or until some data arrives
> > on the connection", which is the more common requirement, of course.
>
> Perhaps an EXPERIMENTAL_ option to send null packets at an interval
> after a specified time period? If the connection doesn't exist, won't
> the null packets fail?
Null packets? On a textual command stream within TCP?
I guess you could try emitting continuation type SMTP error codes - so
if you were going to emit a 200 code you do
while(within_timeout()) {
wait(partial_timeout)
emit("200-We'll get there in the end")
check_status
}
# end of timeout
emit("200 real message")
Its somewhat hacky - and the real arrangement would need to be different
as the most important part of the error message is the first part rather
than the last part.
Is this frankly worth the trouble. Badly behaved people cause a
relatively minor error message. As a DOS it can be controlled with
other restrictions.
Nigel.
--
[ Nigel Metheringham Nigel.Metheringham@??? ]
[ - Comments in this message are my own and not ITO opinion/policy - ]
