From: robert bristow-johnson on
On Apr 7, 6:32 am, Vladimir Vassilevsky <nos...(a)nowhere.com> wrote:
> steveu wrote:
> > Well, there is a theory that people learn from their mistakes, and those
> > hardened by failure are better prepared for the next attempt. I've never
> > seen any evidence to support this. In fact, when most people make exactly
> > the same mistake a second time they can't even see the similarity to the
> > previous occasion.
>
> People learn from their mistakes only if they pay for their mistakes.
> Pain is a good teacher. However, this is not the case if they gamble on
> somebody else's money.

Vlad,

once in a while you say something that i totally agree with, and this
is one of those times.

r b-j

From: robert bristow-johnson on

and BTW, Vlad, may the comp.dsp conference be a rousing success.

sorry i can't make it this or the previous time.

r b-j
From: Jerry Avins on
On 4/6/2010 2:49 PM, glen herrmannsfeldt wrote:
> Jerry Avins<jya(a)ieee.org> wrote:
> (snip on modems and bit rates, where someone wrote)
>
>>> I think one reason for that is over time what constitues a 'phone
>>> line' has also changed. 2400 bps might as well be the limit for a end
>>> to end analog line.
>
>> I don't think so. The unstated assumption behind that line of reasoning
>> is binary signaling (and a signal robust enough not to need error
>> correction). Keep that assumption valid, and the limit really is about 2400.
>
> Well, an important part of going to 9600 was allowing the bands of
> the two directions to overlap. I believe that through 2400 the
> available bandwidth was divided in half, one half for transmit and
> one for receive. (Well, originate and answer in modem terminology.)
>
> In addition, normal (long distance) phone lines use an echo
> surpressor that senses which end is talking and reduces any echo
> from the other end. That is turned off for modem connections.
> At higher rates, where the bands overlap, each has to do echo
> cancelation, detecting a delayed version of its own signal and
> subtracting it.
>
> But yes, and important part of the need for error correction
> is the interference of ones own transmission.
>
> All these are for modems with similar bit rates in each direction.
> There were some simplex (one direction only) and half duplex
> (one direction at a time) modems, and also some non-symmetric
> (people type slower than computers) modems.

In the early days of time-share computing, I used a half-duplex
9600-baud modem and a line-turnaround procedure (request-to-send,
clear-to-send, etc.) When I learned that the modem had a 40-baud reverse
"maintenance" channel. I used that from my TTY to the remote computer,
getting effective full duplex.

Jerry
--
"It does me no injury for my neighbor to say there are 20 gods, or no
God. It neither picks my pocket nor breaks my leg."
Thomas Jefferson to the Virginia House of Delegates in 1776.
���������������������������������������������������������������������