head     1.1;
branch   ;
access   ;
symbols  ;
locks    bin:1.1; strict;
comment  @# @;


1.1
date     91.06.10.10.20.05;  author bin;  state Exp;
branches ;
next     ;


desc
@initial version prov by hal
@



1.1
log
@Initial revision
@
text
@Here are two sets of diffs against the alx.c that I'm currently running.
It was derived from your RTS/CTS-enhanced version (that I dubbed "3.11").

The two sets of diffs are:

- version 3.10 alx.c against my current version
- your enhanced "version 3.11" alx.c against mine.

Important note:  I'm out of my league here.  I'm a reasonably competent
C programmer, but completely ignorant of asynch communications programming
and issues.  Don't look for deep insight in the changes that you see; in
many cases I merely removed code from your enhanced version that I didn't
understand, attempting to narrow down the number of changes involved.

I *have* tested and demonstrated that the code DOES toggle RTS correctly;
the persistent packet errors must be caused by delays in the chain well 
past the driver.

As I mentioned earlier, this version does cure the SSS (Sudden Sluggishness
Syndrome ;-) -- but introduces the following new anomalies:

- My version of ckermit does not get any response from the modem.  The modem
  lights indicate that the modem is sending stuff (e.g. after an ATN?), but
  nothing comes out.  The native plain old kermit does work.

- Recall that I use a fixed-at-9600 line speed, and let the Telebit cope
  with slower modems.  This has worked without incident in the past, but
  no longer.  While uucicos work  (with reduced speed), interactive 
  dial-ups at 2400 hang up the modem as soon as a "large" amount of 
  characters is sent (e.g. "ls -l").  The symptom is that the characters
  come across VERY slowly, and after the output is complete, further 
  keystrokes aren't echoed or responded to by the system.  It's almost as
  if the line speed gets changed from 9600 !?

I look forward to further testing material from you.

@
