BG> I'm running a matched pair of 16Mb 72 pin SIMMs.  The USR anomaly exists in 
BG> every USR model however, and I can think of only one scenario which may 
BG> cause problems - if somebody alters their term's port rate and doesn't 
BG> issue an AT command 

This is not an anomaly, this is they way all modems work.  Auto-baud
only works if you issue an AT command.  If you don't issue one, then
the modem can't use ESP.

BG> (ATZ won't do either), then a call immediately comes in.  

It doesn't have to be immediately.  You might have even dialled
out, finished that, then done an ATZ.  Same problem.

BG> The modem won't answer, as the mailer can't see a RING response in 
BG> order to send an ATA (which, oddly enough WOULD then auto-baud the port 
BG> rate anyway).  With the right mailer init, this problem would NEVER occur, 

ATZ *IS* *A* "right" mailer init.

BG> but it does show how important it is to know that the USRs work differently 
BG> to Rockwells in this regard.

Just as important as it is to know that the Spirit does double
sending when talking to a Rockwell-based modem.

BG> Oh sure, no argument with that at all.  There may well be sinister bugs 
BG> lurking in my USR's EPROM, but if there are, I've not yet found them, but 
BG> as I said earlier, I still don't consider the lack of auto-baud on ATZ to 
BG> be a bug at all.  

Well it is fucked by design then.  I always said that was a
possibility.

BG> Rather, it is a normal reset operation for all Hayes-type 
BG> modems.

Come on, just say it out in full, so that I can get Dave Hatch
to make you look stupid.

BG> Incidentally, I had my first failed connect with his Courier this morning
BG> - a 28800 MNP connect was negotiated, then the modem immediately dropped
BG> carrier.  Called again, got another 28800 MNP connect, and the mail
BG> transferred OK that time.  Most people would blame the modems, but not
BG> me.  Not this time.  I'm not forcing MNP here, so Paul has apparently
BG> been playing around with his Courier's S27 register, something that
BG> should be left well alone unless one knows exactly what they're doing.

db> Yes, he has been playing around .. 

BG> Correct, and he's since admitted as much anyway.

"Admit" is the wrong word.  I ANNOUNCED it.  I also announced when
I stopped it too.

db> but answer this for me - why did the first 28.8Kbps MNP connection fail, 
db> but the second connection of the *same* *type* (ie., 28.8Kbps MNP) *not* 
db> fail? 

BG> Who knows?  Why is Paul's board the only one to which I occasionally have 
BG> problems connecting, regardless of the type of modem in use at his end?  
BG> With different modems, the constant is the software.  I wonder...

What's this "occasional problems connecting"?  You haven't mentioned
that before.  Want to elaborate?  Besides the one where I had
S27=48 anyway.  As for the "constant software" magically making the
USR and Spirits alter their negotiation techniques, you are further
gone than I thought you were Bill.  Even blaming the phone line was
a better attempt.  At least it is on the correct side of the modem
to have an influence.

db> I'm not saying that Paul shouldn't mess with his S27 register, just that 
db> something *else* is at work if one connection is fine and the other isn't.

BG> Quite so, although S27 is not a register to be altered willy-nilly, as it's 
BG> mainly used for incompatibility problems, and while it might fix poor 
BG> connects to one type of modem, it could just as easily screw up previously 
BG> good connects with others.

Pity you never responded to Russell in AUST_MODEMS then, detailing
why I should NOT set S27 as suggested by him.  But then that's you
all over, isn't it?  20/20 Hindsight Bill they used to call you,
eh?

BFN.  Paul.
@EOT:

---