Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!aramis!paul!dpz
From: dpz@paul.UUCP
Newsgroups: comp.dcom.modems
Subject: re: error correction without warning . . .
Message-ID: <257@john.paul.RUTGERS.EDU>
Date: Sun, 8-Mar-87 20:59:17 EST
Article-I.D.: john.257
Posted: Sun Mar  8 20:59:17 1987
Date-Received: Tue, 10-Mar-87 04:23:03 EST
References: <8703021640.AA14087@ucbvax.Berkeley.EDU>
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 46

> From: RANCK@VTVM1.BITNET ("Wm. L. Ranck")

> As Geoff points out there may be no way to tell when your
> modem is doing error correction, although some have lights that indicate
> error correction activity.

Activity... argh - my Racal-Vadic 2400PA's idea of activity is
flashing a light when it is exchanging error correction code, not when
it is actually correcting errors.  Big help - to me that is just as
good as having a light on all of the time signalling enabled error
correction, and if you have this modem rack mounted, it is all but
ignored in daily use.  To have it flash when a correction actually
takes place would be more helpful, but then again, staring at my
modem, waiting for it to flash a correction, isn't very high on ways
that I like to spend my time.  Some kind of internal log of actual
error correction inside of the modem would be nice.  This way you
could check after a file transfer what happened inside that little
black box.  What might even be better is for the modem level xfer
protocol (terminal/file xfer program in the case of a micro) to get
together with the modem and relay error messages up higher.  Ad
{nauseam, infinitum}.

>   Then there is the question of error correction at the various levels
> in the process.  If your software does error correction and your modem
> does error correction and the network does error correction, each layer
> adding a few more bytes, you end up with more bytes of error
> detection/correction than data, and of course more bytes means more chance
> of error.  In some cases, though I can't think of any, the error correcting
> protocols might interfere with each other.

Strength (stench?) in numbers! :-) This is, of course, a trade-off
that we make for ensuring the 'rightness' of our data.  The problem
with all of this is that each is independant, so you can use an error
correcting modem without an error correcting network (OK, without a
network, period), and you can use an error correcting network with a
non-error-correcting modem.  *sigh* What I need is a nice toggle
switch to turn the correcting on or off for each layer, and leave it
on only for those layers that actually provide the 'best' correction
algorithms.

						dpz
-- 
David P. Zimmerman	"When I'm having fun, the world doesn't exist."

Arpa: dpz@rutgers.edu
Uucp: ...{harvard | seismo | pyramid}!rutgers!dpz