> On Mon, 25 Aug 1997, Tom Samplonius wrote:
>
> > This is bogus. A all-zero async map means that no characters are
> >escaped, so everything is transmitted as-is. So, so all those
> >"special characters" are still being transmitted over the link. Besides
> >"special characters" aren't supposed to cause retrains, because the
> >connection should be 8-bit clean.
>
> This is NOT bogus if the modems try to interpret a specific sequence of
> characters as some kind of command, akin to sending a "+++" if that escape
> to command mode hasn't been disabled. The connections can be 8 bit clean
"+++" requires a pause after it, in order to enable command mode. If
that occurs in a stream of PPP packets, then your PPP stack has big
problems.
> with or without escaping certain characters. We are talking about certain
> control characters, not the presence/absence of the high order bit. The
> control characters are passed in either case, but in the case of escaping
> them, a two byte substitution is made for the offending control character.
Doesn't matter. "8-bit clean" means that all combinations of 8 bits can
pass through the channel without modification.
> The example is the previous example given... The escaping of xon/xoff.
> If your modem is not set properly, it will listen to xon/xoff.. With the
> characters escaped, the modem does not see the xon/xoff but the
> connections on the other end of the ppp connection do. Besides, how do
> you explain the several reports of disconnects while in ppp mode vice
> none in plain telnet? And this is with the same equipment on both ends
> as to modems, computers, term servers, etc.
Those reports were never proved. In both cases it was mentioned that
terminal access OR ppp was the only thing tested. Besides terminal
traffic is not a stressful as FTP download over PPP, that is terminal
traffic is unlikely to run the channel at 100% utilization.
Tom