[Oberon] Repost bounce ? [Oberon]

[eas-lab at absamail.co.za]

Received: from exim by bernie.ethz.ch with local (Exim 4.05)

A message that you sent could not be delivered to one or more of its
recipients. This is a permanent error. The following address(es) failed:

  
------ This is a copy of the message, [snip...] ------

	by absamail.co.za with NetMail SMTP Agent $Revision:   
3.14  $ on Novell NetWare;
	Fri, 13 Jun 2003 14:42:02 +0200
From:  eas-lab at absamail.co.za
X-Mailer: Oberon Mail (ejz) on PC Native 11.10.2001
To: oberon at inf.ethz.ch
Cc: eas-lab at absamail.co.za
Subject: Re: [Oberon] Native: PPPMain + tasking + Ctrl/Break ?
Message-Id: <E19QnoU-0006iJ-00 at bernie.ethz.ch>
Date: Fri, 13 Jun 2003 14:37:51 +0200

> > when the ppp 'connection' is unsuccesfull, the next dialup attempt
> > traces: --------
> > 
> > Dial script startedModem off-line, device 0 closed
> > 
> > Sending [ATZ]
> > Waiting 10s for [OK] {
> >  --- and a few seconds pause, then
> 10 seconds probably :-)
> 
> > }
> > Dialer: Timed out
> > Dialer: Script aborted
> 
> Edgar wrote:
> This appears to me like the modem is in a bad state and isn't reacting on
> an ATZ. Perhaps because it has still some connection to the switch.
> Do you try to turn off/on the modem power in this case ?
> 

I remember the TV repairman who told that when he was called to the
old-age-home, he FISRT took the back-cover off, before he checked
that they had NOT switched it on at the wall.    How else could he
justify his call out fee ?

This is the 'level' to which this 'investigation' has gone:--
System.Directory Devlp:PPPanaly\d == 11.06.2003 20:32:48  371022

And file Devlp:PPPanaly starts like this:-
2. Peter Easthope's Home page
3. mailing list posts
4. PPP Notes - http://carnot.pathology.ubc.ca/PPPnotes.html
5. Log of RFCs fetched
6. Book headings: PPP for Embedded Systems
7. Extract from NtwrkAdmDocm - linux
8.  PROCEDURE CloseCOM2*;  vs. PROCEDURE Remove*(VAR id:PPPid);
9. trace of error symptoms.
10. Analyse NewsGrp traces
.....
16. reset-serial /dev/ttyS1
17. HDLC summary
18. Ctrl/Break - Oberon.Loop
19. Use V24.Panel at peer to monitor output.
20. Suspect DTE
21. http://axion.physics.ubc.ca/ppp-linux.html
22. URI encoding - O-URL-code.html
23. XON/XOFF

Ie. the file of 371K now has 23 'chapters' !!
The index looks much better in colour.
I spend half an hour a week average investigating this 'dog'.

As I recall the condition applied also on my other scrap-box.
The last idea was to send hex11 to the modem (the default XON?XOFF).
But I can't because (as has been disclosed during this investigation)
V24 can't be run until ppp releases (apparently the port).
And if I Free ?pppMain? as indicated in my previous post,
I need to reboot, and the problem is gone.

And now I remember that I have recorded the various status register
values of the ser. chip, when in and out of the 'problem-state'.
I.e. the problem occurs before the modem - so XON/XOFF would
have no effect. Or does the 'ready' line(s) from the modem
effect the ser.chip status ?  Probably.  But probably I checked this by
removing the modem ?

This is the kind of 'exotic' thinking that is required:---
  " The other box had a period when (depending on the weather)
n-o whould not get dial tone, although the same hardware under
DOS and linux would.  This suggests a timing issue. 
My second modem required extra " ATX" to recognise dial tone.
This is quiet common in s. africa - apparently the telco conditions."

The person who can immediately see the cause of the current bug 
is the person who understands the multi-tasking of n-o and how the 
Ctrl/Break effects things, down at the machine code level."

-- Chris Glur
----------------  addition to second post:-
I've:
  * conected another machine with V24 in place of the modem;
   it looks OK except that the <line terminator> is 'problematic'
    with V24. I think modems want CR and n-o V24 wants LF.

 * confirmded that the ser-port status regs are influenced by
    switching the modem on/off.   It's a FSM, in that the ser-port
   status regs are changed by READING.
----
It's time to look beyond simple explanations: think of multi-tasking
effects.   Why does Keyboard: Ctrl/Break 'unlock' the problem ?