Xref: utzoo comp.dcom.modems:3014 comp.sys.att:4811
Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!rutgers!cbmvax!ditto
From: ditto@cbmvax.UUCP (Michael "Ford" Ditto)
Newsgroups: comp.dcom.modems,comp.sys.att
Subject: Re: Verbose modems (Re: MORE 6386 UUCP WOES)
Summary: "last hop" data rate doesn't mean much
Keywords: modem baud rate
Message-ID: <5319@cbmvax.UUCP>
Date: 26 Nov 88 19:00:54 GMT
References: <319@argon.UUCP> <2096@cuuxb.ATT.COM> <727@wsccs.UUCP> <889@vsi.COM> <758@wsccs.UUCP> <83@prapc2.UUCP> <2174@cuuxb.ATT.COM> <5203@cbmvax.UUCP> <6916@chinet.chi.il.us> <5212@cbmvax.UUCP> <1988Nov24.004706.7463@ziebmef.uucp>
Reply-To: ditto@cbmvax.UUCP (Michael "Ford" Ditto)
Organization: Commodore Technology, West Chester, PA
Lines: 39

In article <1988Nov24.004706.7463@ziebmef.uucp> cks@ziebmef.UUCP (Chris Siebenmann) writes:
[ in reply to discussion about the computer knowing the "real" baud rate of
  a modem connection ]
>
> The problem with this is that at different baud rates the user may
>wish drastically different behavior from programs,

This is true, but is a more general problem than the modem issue.  Such
programs must base their idea of the user's communication rate on something
more than the rate of the "last hop" of the connection.

Consider this example:  I am running a terminal program on my Amiga which
can display characters at a nominal rate equivalent to 9600 bps.  It is
communicating via a serial link at 19200 baud to a Unix machine with a
built-in 1200 baud modem.  The modem places a call to a terminal server
which connects via ethernet (10Mbps) to a Unix machine which believes the
connection is at 38400 baud.  Now, does it make any difference whether the
answering modem talks to its host (the terminal server) at 1200, 300, or
38400 baud?  And what about the on-board modem that will speak 1200bps to
the phone line regardless of whether the port is ioctl'd to 1200 or 9600
baud.  Added to all the other unknowns is the existence of devices with
data-dependent throughput, like modems with data compression.

I think the best that can be hoped for in this (increasingly common) type of
situation is for all the communication links to be as "dumb" as possible and
just provide a reliable, flow controlled 8-bit connection, and for the "host"
to be directly told what rate to "pretend" the connection has (1200bps in
the example).

Of course, if it were practical for the host to automatically figure out
the rate of the "bottleneck point" of the connection, that might be a
good idea.
-- 
					-=] Ford [=-

"The number of Unix installations	(In Real Life:  Mike Ditto)
has grown to 10, with more expected."	ford@kenobi.cts.com
- The Unix Programmer's Manual,		...!sdcsvax!crash!elgar!ford
  2nd Edition, June, 1972.		ditto@cbmvax.commodore.com