home *** CD-ROM | disk | FTP | other *** search
- TCP-Group Digest Tue, 25 Sep 90 Volume 90 : Issue 148
-
- Today's Topics:
- Anders can you hear me?
- For *real* network weenies only ...
- FORTH under NOS
- NOS under Windows
- PPP protocol info (3 msgs)
- Servers not starting in NOS
-
- Send Replies or notes for publication to: <TCP-Group@UCSD.Edu>.
- Subscription requests to <TCP-Group-REQUEST@UCSD.Edu>.
- Problems you can't solve otherwise to brian@ucsd.edu.
-
- Archives of past issues of the TCP-Group Digest are available
- (by FTP only) from UCSD.Edu in directory "mailarchives".
- ----------------------------------------------------------------------
-
- Date: Mon, 24 Sep 90 6:41:50 PDT
- From: <marc@kg6kf.AMPR.ORG>
- Subject: Anders can you hear me?
- To: tcp-group@ucsd.edu
-
- Anders,
- I have tried several times to send you e-mail, to klemets@sics.se.
- Is it not getting through?
- I want to discuss NOS Forth with you. Please try me at the following:
-
- marc@toaster.sfsu.edu
- mdg@postgres.berkeley.edu
- marc@toad.com
- marc@well.sf.ca.us
-
- Thank you!
-
- And thanks to the group for indulging this slightly inappropriate use
- of the bandwidth.
-
- Marc de Groot KG6KF
-
- ------------------------------
-
- Date: Tue, 25 Sep 90 09:42:04 +0200
- From: klemets@sics.se
- Subject: For *real* network weenies only ...
- To: collinge@uvicctr.uvic.ca
-
- > What is your favourite network book I can look it up in?
-
- Try "Data Networks" by D. Bertsekas and R. Gallagher, Prentice-Hall 1987.
- That is a good textbook on throughput analysis, routing, etc.
-
- Anders
-
- ------------------------------
-
- Date: Mon, 24 Sep 90 21:05:13 UTC
- From: iw0cnb@tomcat.gsfc.nasa.gov
- Subject: FORTH under NOS
- To: tcp-group@ucsd.edu
-
- Nice to hear about new version, I tried the first one and it's great!
- I made the command "forth" to open a new session so I can open multiple forth
- interpreters. But there are memory problems, I often get "freeing garbage"
- with associated NOS crash...
-
- Will it be possible to interface it with the TCP sockets so we can write NOS
- applications in FORTH?????
-
- Max
-
- iw0cnb@tomcat.gfsc.nasa.gov
-
- ------------------------------
-
- Date: Mon, 24 Sep 90 21:13:20 UTC
- From: iw0cnb@tomcat.gsfc.nasa.gov
- Subject: NOS under Windows
- To: tcp-group@ucsd.edu
-
- For those who might care, I use the following PIF settings for NOS under
- Windows 3.0 in 386 mode:
-
- Memory used: 400 kb
- Memory required: 640 kb
- Full-screen and no background (I can change them later, in the "settings" window
- In the advanced options, I turned off EMS usage and High memory.
- Since we work in text mode, I unchecked all the boxes regarding graphics.
- No text mode emulation and no video memory retain.
-
- As to the background and foreground priority, it works ok with the default
- values of 50 and 100. I tried to give it maximum priority, and it almost works
- ok at 4800 while in text mode, but as soon as I virtualize the screen (put it
- in a window), it loses characters.
-
- I also put in the system.ini file the line COM1BUFFER=1024, but it doesn't seem
- to help much. Maybe we should make NOS aware of Windows, like it is with
- DesqView and DoubleDos, where it doesn't lose characters. Any idea is welcome...
-
- Max
-
- iw0cnb@tomcat.gsfc.nasa.gov
-
- ------------------------------
-
- Date: Mon, 24 Sep 90 21:24:09 UTC
- From: iw0cnb@tomcat.gsfc.nasa.gov
- Subject: PPP protocol info
- To: tcp-group@ucsd.edu
-
- I downloaded PA0GRI sources which include the Point to Point protocol.
- Since there were no doc, I tried to grasp something from the source comments.
- I understand that this protocol (and the associated compression routines)
- is intended for slip work. Is it possible to use it in a packet radio
- environment, possibly for SMTP compressed forwarding???
-
- We are trying to put up a TCP/IP HF link using 1200 PSK, so data transfer
- time should be kept low....
-
- Max
-
- iw0cnb@tomcat.gsfc.nasa.gov
-
- ------------------------------
-
- Date: Mon, 24 Sep 90 17:32:11 PDT
- From: Brian Lloyd <brian@robin.telebit.COM>
- Subject: PPP protocol info
- To: tcp-group@ucsd.edu
-
- No, PPP is not designed in such a way that it can be used over the
- radio. The basic PPP frame is an HDLC frame with the address field
- set to ff, the control field set to 03 (UI), a two octet type field,
- and then the data which is usually an IP dgram. Because of the
- address field, ctl field, and the two octet type field it is less
- efficient than SLIP (I guess that it is still more efficient than
- AX.25 but that is another story :-).
-
- The one item that PPP supports that would be a win for AX.25 (and
- could be applied to AX.25 with little or no effort) is an option
- negotiation process that allows you to negotiate away just about
- everything (like the useless address and control fields and the first
- octet of the type field).
-
- One of the items that it does allow you to negotiate ON is Van
- Jacobson's TCP/IP header compression. Basically Van's compression
- allows the receiver to predict what the next header will look like and
- thus allow the sender to send only that information that really
- changes from packet to packet. The result for TELNET, FTP, and SMTP
- sessions is that most of the TCP+IP headers gets squashed to 5 or 6
- octets instead of the usual 40 octets. This would be a big win for HF
- use. You could drop your packet size way down and not take the huge
- throughput hit.
-
- You don't get something for nothing though. The first TCP+IP header
- must get through uncompressed: all 40 octets of it. The sender
- basically sends header deltas from there. Bottom line is that your
- link must be good enough to get the first packet through.
-
- Maybe someone ought to offer up an extension to AX.25 that includes
- option negotiation and header prediction/compression.
-
- 73 de Brian, WB6RQN
-
- ------------------------------
-
- Date: Mon, 24 Sep 90 22:20:14 -0400
- From: "Louis A. Mamakos" <louie@sayshell.umd.edu>
- Subject: PPP protocol info
- To: tcp-group@ucsd.edu
-
- An assumption made in VJ TCP header compression algorithms is that the
- link is somewhat reliable; at the slightest provocation, the alogrithm
- will revert to sending a complete, uncompressed TCP segment. This is
- usually the result when the TCP retransmits. It's not all roses.
-
- If you were to use header compression, you'd might think about running it
- over an connected mode AX.25 circuit. Though, you would have to think about
- how efficient that is compared to TCP's ability to reassembly out of order
- segments and its generally better retransmission strategies.
-
- Or maybe you could use the effective added bandwidth gained by header
- compression to include some FEC "overhead"?
-
- louie
- WA3YMH
-
- ------------------------------
-
- Date: Mon, 24 Sep 90 11:38:00 EDT
- From: Marcus (M.D.) Leech <MLEECH@BNR.CA>
- Subject: Servers not starting in NOS
- To: tcp-group@ucsd.edu
-
- We're currently having a problem with NOS0828 with the servers not
- starting. Mostly it's the TELNET server, but sometimes other
- servers won't start (FINGER, ECHO, DISCARD).
-
- When you go to start them manually, you have to "stop <servername>" even
- though there's no process-table entry or socket-table entry associated
- with the process. It's almost like the process gets "half-started" and
- then silently dies leaving enough cruft around that you have to
- stop it before you can restart it. Has anyone else experienced this
- problem? When this problems happens, there's no obvious memory
- starvation going on. I've tried starting with the interfaces unplugged, to
- avoid any possible interrupt-caused-races within the kernel--it still
- happens when I do that.
-
- --
- -----------------
- Marcus Leech, 4Y11 Bell-Northern Research |opinions expressed
- mleech@bnr.ca P.O. Box 3511, Stn. C |are my own, and not
- VE3MDL@VE3JF.ON.CAN.NA Ottawa, ON, CANADA |necessarily BNRs
-
- ------------------------------
-
- End of TCP-Group Digest
- ******************************