[Oberon] monitoring an RS-232 interface.

[eas-lab at absamail.co.za]

peter_easthope wrote:
> Here is a rather unusual problem.

Answers to my NewsGroup queries described such a device as 
being essential. It's a worth while project.

> I want to monitor the serial interface
> of an external modem.  The approach I am 
> considering is a generalization of that 
> described in 
> "http://www.oberon.ethz.ch/native/WebTrace.html" .

 I've archived this. It's only about 10 lines. Is that right ?
 "WebTrace" seems to be a poorly chosen name.

> In this case we need to monitor both sending and 
> receiving lines.  This will require a tee serial cable with 
> the side branch dividing to two cables.  One of these 
> side branches needs the send and receive lines 
> crossed.  This special cable will be inserted between 
> the modem and the server.  Then one "observer" 
> machine can be attached to each of the side branches.
> 
Yes: 
Tx ser-port = Rx modem --> Rx monitor 1
Rx ser-port = Tx modem --> Rx monitor 2

> WIth this setup, one observer will monitor transmissions 
> from computer to modem and the other will monitor 
> transmissions from modem to computer.  Perhaps one 
> machine with two serial ports is capable of monitoring 
> both lines.  No doubt this would work with Unix but I 
> have not tried running two concurrent V24 tasks in Oberon.
> 
The serial port(s) only need attention when their FIFO (buffers)
are getting full or empty for Rx and Tx respectively.

Has n-o got the ability to interrupt for modem or mouse ?
I don't think so. Older machines had smaller buffered ser-chips
and would be more likely to 'overflow'. 

Normally, the ser-mouse and ser-modem ARE 'sharing' tasks.
Although not symetrically. From the papers which you kindly posted:
" Mouse and keyboard buffers are polled at each task switch in the 
loop.."

The whole business of how (normally) the ser-modem is given and
releases its ser-port, is central to the ppp-bug which I'm 
investigating, and where readers 'look the other way' when
I ask questions. =:-(

> Is such a technique described elsewhere?  Any suggestions
> or improvements on my proposal?

There's a special name for this tool (which exists) that was 
mentioned in 2 of my threads in Newsgroups: comp.protocols.ppp.
Perhaps a google will lead to mention of how it works.
I think your idea is correct/do-able.
You could start (a-la Wirth) with a minimal system, by using your
modem port to monitor your mouse port.  This will prove to you 
if/how the basic concept is OK.

What is the serial port speed setting for the mouse ?

Chris Glur 

PS. keep me informed of progress please.

Thomas wrote:

> There is such a program for Aos/Bluebottle. You can link the watch 
> computer between
> the modem and the control computer. It usually works well as long 
> as the connection
> speed is not changed during the connection.
> I'll have to search for the program if you are interested.

Does it monitor both control computer AND modem.