BG> Ask Rod to disable retrains at his end, not yours.

RS> Its not MY modem asking for a retrain within a second of the
RS> initial connect on every single call. Thats never seen with
RS> anything except a USR, so its got to be the USR doing that.

BG> Likely so, but I was curious to see what would
BG> happen if you disabled retrains at your end.

I'm not convinced I can in the sense of refusing retrains the USR demands.
The Supra docs seem to be pretty unambiguously stating that you can only
control the Supras monitoring of the line quality and using that to
initiate either a retraining or a rate renegotiation, fallback/forward.

BG> Does the call continue at 14k4, or does the Courier drop DTR?

On the good calls, the call continues at 14400 and never ever changes
speed from 14400. On the dud calls, its STILL at 14400 after the retrain,
but with continuous receive and transmit error retrys between the modems
at the LAPM level, and eventually my mailer gives up because it cant get
the session handshake at the mailer level thru that. On one occasions only
it did eventually success, taking heaps longer to mailer handshake than usual.

RS> The main question mark at the moment is whether
RS> that was just a wart in earlier USR flashrom code.

BG> It's a few SDLs back, but there were some problems
BG> with trigger-happy retrains, as far as I can recall.
BG> I would hope that's been fixed by now.

It doesnt appear to be that coz it just mindlessly demands
a retrain VERY soon after the CONNECT and NEVER EVER asks
for another during the entire session. So it doesnt appear
to be just too trigger happy on the line quality.

PE> 5. Paul Markham's Netcomm - connections at 26400, 24000, 2400!, 14400.

BG> Lines again?  Dunno, it's extremely unusual to say the least.

RS> Cant be lines if Paul Markham doesnt get
RS> that sort of behaviour calling other systems.

BG> I mean the lines between PM and PE.

Still cant see it given he could call the Spirit ok.

BG> Given my own long-term results, and bearing in mind that
BG> everywhere else I call is at 28800 straight up, I suspect
BG> that there could be a line problem at PE's end.

Clutching as straws Bill, particularly with
a 2400 connect. Dont believe it. 14400 either.

PE> to a wide variety of common modems, including all of
PE> the above. The USR Courier is falling way short of that.

BG> Shouldn't be,

RS> But clear is. And you shouldnt get such gross problems, particularly
RS> the 2400 connects, with a very basic modem config from factory.

BG> Dunno if you read the USR_MODEMS echo, but I know you read AUST_MODEMS,
BG> and nobody else has EVER reported similar problems with the Courier.

I've seen that over and over again, one modem looks like gods gift to
modems, it turn out not to be. And Poe has had some significant problems.

And Paul Markham has had problems calling OUT on HIS line with it too.

BG> Maybe Paul's place is jinxed...

Or maybe the sun doesnt actually shine out of the USR VE's arse |-)

--- PQWK202