BG> Did my Sportster demand a retrain when you called it the other day ?

RS> Nope, only Couriers do, and only when the V32terbo is
RS> enabled. Which does sort of hint at some quirk in the USR
RS> V32terbo code since the Sportster doesnt support that mode.

BG> Could it also be something to do with the fact that
BG> USR have proprietarily extended V.32ter to 21600,

Yes, thats certainly quite possible. I couldnt see any
obvious way to disable JUST the 21600 speed and not the
19200 V32terbo, but then I didnt look that hard I must admit.

BG> and it may be attempting to train upwards
BG> following the initial 19200 connect ?

No, its always gets a perfect 14400 connect every time and its
appears to be attempting a retrain to see if 19200 is feasible.
Corse its quite possible that if the Courier could be told to
not attempt 21600 V32terbo, that that might work fine. If its
possible to actually tell a Courier to do that.

RS> There is a sort of rather warped logic there I suppose. As far as I
RS> can recall you cant fall forward from V32bis to V32terbo, you do need
RS> to do a retrain, and the 14400 achieved on the connect is certainly
RS> below the V32terbo speed. So it appears that even when the Courier has
RS> got a perfect V32bis connect every time, it just mindlessly demands an
RS> attempt to get V32terbo, even tho the other end doesnt support that.

BG> Or it may even be related to USR's proprietary V.32bis
"fast-connect",

We tried disabling both of those, S34.2=1 and S34.128=1, tried
individually. Didnt actually try both at once come to think of it.

BG> which, I might add, works extremely well
BG> (as long as you don't own an old Supra).  :)

It bit quite a few existing modems back in the days when USR
first started pushing it with V32bis. And since it causes the
Supra no problems with a Sportster, it cant be that alone.

RS> Its less clear why Ian doesnt see that
RS> behaviour with his DPX V32bis calls tho.

BG> To date, you're the only person who has reported this "problem".

Trouble with that line of argument is that I've heard it all
before Bill. Thats what everyone said with the Spirit double
send, and it turned out that heaps of people were actually seeing
it and many didnt even realise that there was a problem at all.

Same with the USR ATZ for that matter, its been there for YEARS and
just requires a particular combination of circumstances to bite you.

Thats the reason why it makes a hell of a lot of sense
to minimise the number of gotchas in a particular
design, coz they will inevitably bite sometime.

BG> I wonder if you've bothered trying any other modems at all.
BG> The results would be extremely interesting, to say the least...

Its a bit hard. Most dont even show that they are seeing a retrain
demand from the other end, and its not a trivial matter to get someone
who isnt doesnt know a bit about modems to try it and report back either.
@EOT:

---