home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
kermit.columbia.edu
/
kermit.columbia.edu.tar
/
kermit.columbia.edu
/
newsgroups
/
misc.19950726-19950929
/
000366_news@columbia.edu_Fri Sep 15 05:30:37 1995.msg
< prev
next >
Wrap
Internet Message Format
|
1995-12-25
|
4KB
Received: from apakabar.cc.columbia.edu by watsun.cc.columbia.edu with SMTP id AA16591
(5.65c+CU/IDA-1.4.4/HLK for <kermit.misc@watsun.cc.columbia.edu>); Fri, 15 Sep 1995 22:02:55 -0400
Received: by apakabar.cc.columbia.edu id AA23706
(5.65c+CU/IDA-1.4.4/HLK for kermit.misc@watsun); Fri, 15 Sep 1995 22:02:54 -0400
Path: news.columbia.edu!panix!bloom-beacon.mit.edu!paperboy.osf.org!bone.think.com!blanket.mitre.org!agate!howland.reston.ans.net!cs.utexas.edu!news.cs.utah.edu!cc.usu.edu!jrd
From: jrd@cc.usu.edu (Joe Doupnik)
Newsgroups: comp.protocols.kermit.misc,comp.protocols.tcp-ip.ibmpc
Subject: Re: Problems with MS-Kermit 3.14 and (c)slipper.exe?
Message-Id: <1995Sep15.113038.61385@cc.usu.edu>
Date: 15 Sep 95 11:30:37 MDT
References: <43btto$g66@apakabar.cc.columbia.edu>
Organization: Utah State University
Lines: 52
Xref: news.columbia.edu comp.protocols.kermit.misc:3632 comp.protocols.tcp-ip.ibmpc:37069
Apparently-To: kermit.misc@watsun.cc.columbia.edu
In article <43btto$g66@apakabar.cc.columbia.edu>, ylee@manila.cc.columbia.edu (Yeechang Lee) writes:
> I use Kermit 3.14 (24 May 1995 release, patchlevel 9) and slip8250.com
> (10 Feb 1992, the one that comes with the Kermit .ZIP distribution)
> quite successfully on my 386sx/16 to dial up to Columbia and run
> multiple terminal sessions with SLIP. However, running Kermit with
> either slipper.exe or cslipper.exe (15 Jan 1994, part of slippr15.zip
> file available everywhere--their documentation describe them as
> drop-in replacements for slip8250.com) is not so successful--about
> half my keystrokes get "bunched up" for several seconds at a time,
> then get all spit out to the host at once. Output to me seems
> unaffected.
This looks rather like lost packets, going in either direction.
The 24 May edition of MSK has slight changes in the transmit section
where Nagle's algorithm is involved (suspend further transmission until
an outstanding one has been ACK'd, that kind of thing).
> The funny thing is the trouble didn't exist with the 18 January Kermit
> release; I used cslipper.exe and Kermit on an old 8-MHz Tandy clone
> quite successfully. Upgrading to 24 May caused similar symptons on
> that machine (verified by returning to the older kermit.exe
> executable). Now same things happen on this 386sx/16. Any ideas?
The older MSK was more aggressive about sending, the newer one
is closer to industry standard practice on the Nagle matter. Differences
will arise when outgoing data or their ACKs are lost. The newer one is
preferable overall.
> Here are the parameters I use to load slip8250.com:
>
> c:\kermit\networks\slip8250 0x60 -h slip 3 0x2f8 38400
>
> And slipper.exe and cslipper.exe:
>
> c:\kermit\networks\slipper.exe com2 baud=38400 keepalive
>
> I have tried loading slipper.exe and cslipper.exe in regular memory,
> in upper memory, and with and without the keepalive switch, all with
> no effect.
I have to guess that maybe flow control to the modems is not all
that it ought to be, or the modem is buffering material and sitting on it.
Doing VJ header compression yields shorter packets and hence the modem
buffer fills more slowly.
Using slipper.exe here on a straight copper wire across the hall
way has performed ok.
There is one more remote possibility, and that is maybe (c)slipper
tries to be cute and use TX empty interrupts from the UART to trigger more
sending. That can easily lead to loss of received-char-available interrupts
and hence lost incoming packets. That's the reason I rewrote slip8250 some
time ago.
Joe D.