Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!oliveb!tymix!antares!pnelson
From: pnelson@antares.UUCP (Phil Nelson)
Newsgroups: comp.sys.amiga
Subject: X-off and A-Talk "feature"
Summary: A suggestion as to when not to obey protocol
Message-ID: <382@antares.UUCP>
Date: 8 Feb 89 08:07:05 GMT
Reply-To: pnelson@antares.UUCP (Phil Nelson)
Organization: Tymnet QSATS, San Jose CA
Lines: 46


 Well, I still think Marco writes a good Comm program, but I have found
a "feature" (not a bug exactly, but it sure bit like one). The problem has
to do with how you deal with the receipt of a stray x-off (control-s)
character. I found out (when I received a stray x-off) that your program
ceases output. Now I know that this is what it is supposed to do, and some
other programs do the same, but consider; if I never receive the x-on, I
can never send any data again. In fact, I could not even exit the program,
I had to re-boot the Amiga. Naturally, all this occurred at a very bad time,
so I simply went through my terminal program collection until I found one
that worked (well, to be fair, one that didn't have this feature), vt100.
I would like to take this opportunity to thank Dave Wecker and Tony Sumrall
for a very useful program. I also tried the Gizmoz Terminal, which had the
same feature.

 I realize (after dragging out the datascope to discover the cause of the
problems I was having, I didn't know it at the time) that I can simply
disable the X-on - X-off, but I normally run with it enabled, for those
cases where I wish to upload files straight (without protocol). I can think
of three other ways to handle this;

1. Ignore the receipt of X-off when data is coming from the keyboard (I think
   this is what vt100 does, it's also the way I did it when I wrote a comm
   program once).

2. Timeout - MSKermit does something like this (I encountered the same
   embedded X-Off at work, running MSKermit (also a very useful program,
   for those of us who have to use ..shudder.. clones :-). MSKermit handled
   the same situation by considering the matter for about 10 seconds after
   I started typing, then, finally, sending the character. I wasn't really
   pleased with this, but it did let me continue working.

3. Alert the user, and provide some means of override - I haven't seen this,
   but it seems like a reasonable compromise for those who do not want to
   ignore the X-off under any circumstances. As I see it, if the program is
   not going to send my characters (it's normal function), it should tell
   me, at least.


So, I am still using A-Talk III, except for this one problem it has served
very well, but I keep vt100 around for emergencies.
-- 
Phil Nelson at (but not speaking for)                  OnTyme:NSC.P/Nelson
Tymnet, McDonnell Douglas Network Systems Company       Voice:408-922-7508
UUCP:{pyramid|ames}oliveb!tymix!pnelson              LRV:Component Station
"ding ding..." -Santa Clara County Transit Company trolley car (AKA "LRV")