PE> One solution to this problem would be for me to go back to my old method,
PE> and you guys all setting your modems to expire after about 3 minutes or
PE> something, whatever it is before Telecom gives up on your behalf.

RS> Thats a bit tricky coz there can be significant variation in the
RS> handshaking time, particularly with some particular circumstances.

PE> Pardon?  It should work fine.

RS> Not when the handshaking time in successful sessions varys extensively
RS> as you change modems at your end. Particularly when the time that
RS> matters is the longest time thats ever seen for a successful
RS> handshake, plus a margin for safety. The longer that gets, and it has
RS> to be reasonably long, the more likely it is for you to come back on
RS> line late in the dial attempt, loop the line coz its ringing, and then
RS> the caller times out coz there is nothing like enough time left to
RS> successfully handshake. In other words the callers system aborts the
RS> handshake coz you looped the line very late in its timed event.

PE> My idea is for the callers to set their timeout to about 4 minutes.

I cant see what you think that will achieve,
thats no different to the Telstra timeout.

And when its that long, you have done nothing different
to what you had before you busied the modem.

RS> Maybe its possible to do it your way, with the modem not being
RS> busied, and then have a special event which is invoked at the end of
RS> the mail processing event which just checks to see if the modem is
RS> seeing a ring, and spins its wheels till it stops ringing, and doesnt
RS> answer the call. Then once the line has stopped ringing, that event
RS> terminates and you drop into the normal mode waiting for the next RING.

PE> Yes, that sounds good, although you may have trouble
PE> with one person aborting and the next person getting
PE> through, so that it takes ages for the RINGs to clear,

I think you are having another brain fart. This proposal is no different
to the situation where you just dont answer the phone during the mail
processing, and just avoids the risk of picking up the call at the end
of a long series of rings, just before the caller is about to give up,
only to have his mailer timeout very soon after you answer.

It has no effect whatever on the separate
problem of more than one caller etc.

PE> although that could be countered with a 5 minute timeout.

I think you are misunderstanding what I intended there.
@EOT:

---