FidoNews · Vol 2, No 12 · 6 May 1985
FIDONEWS -- 06 May 85 00:00:48 Page 1
Volume 2, Number 12 6 May 1985
+----------------------------------------------------------+
| _ |
| / \ |
| - FidoNews - /|oo \ |
| (_| /_) |
| Fido and FidoNet _`@/_ \ _ |
| Users Group | | \ \\ |
| Newsletter | (*) | \ )) |
| ______ |__U__| / \// |
| / FIDO \ _//|| _\ / |
| (________) (_/(_|(____/ |
| (jm) |
+----------------------------------------------------------+
Publisher: Fido #375
Chief Procrastinator: Thom Henderson
Fidonews is published weekly by SEAboard, Fido 107/375. You
are encouraged to submit articles for publication in
Fidonews. Article submission standards are contained in the
file FIDONEWS.DOC, available from Fido 107/375.
Disclaimer or don't-blame-us:
The contents of the articles contained here are not our
responsibility, nor do we necessarily agree with them;
everything here is subject to debate. We publish EVERYTHING
received.
Any Day Now
The hot question on Fidonet these days is "When do I get
10h?" We now have the definitive answer; you don't. It
seems that there have been several versions of 10h (?), so
it's going to be called 10i when it is officially released,
in order to ensure that everyone has the same version.
So when do we get our hands on 10i? The word I got last
Wednesday was that it would be released no later than last
Friday. So what happened? Well, it seems that Tom Jennings
thought of another feature to add at the last minute, so
release has been delayed until the new (newer?) version is
fully tested.
The new feature is a goodie, and should make many peoples'
lives much easier, so it's worth the wait. It's the ability
to have one Fido be two different net/node numbers (such as
being, say, 1230/1 on a private net, while still being
107/375 on the public net). In the new scheme of things, we
will be node 375 in net 107 (Metro NY), but I plan on
setting 1/375 as my alternate net/node for awhile until I'm
pretty sure everyone has the new stuff working. What you
FIDONEWS -- 06 May 85 00:00:51 Page 2
will do, I have no idea.
One option you DON'T have is to leave things alone. They're
serious about this, guys. The official rumor I hear is that
anyone who doesn't convert to the new setup within a month
or two is going to get dropped from the national list. The
guys in St. Louis have been doing a bang-up job, but it's
just gotten too big for mere mortals to cope with.
NODELIST and ROUTEGEN will also have to change, but there's
a bit of a hitch there. It seems John Warren (the guy who
wrote them) is in the hospital. I'm told he's been bugging
his wife to bring in his computer (a sentiment I can well
understand), but I wouldn't count on anything for awhile.
Let's just leave the guy in peace, to get well at his own
pace. Good luck to you, John!
So when do we get 10i? Any Day Now. In fact, this
editorial is in severe danger of being rendered obsolete
even as I write it (Saturday night). It's to be released
just as soon as Tom Jennings feels secure about it. From
what I've seen, that won't be too long. I've test-run a
copy of 10h, and I can testify that it looks pretty good.
If it has any bugs, I didn't find them.
There IS one small thing I found, though. This is just a
little unofficial advice from me to you. Don't bring up 10i
while you still have pending outgoing messages. In my case,
I had a whole bunch of mail waiting to go out when I brought
up 10h and told it I was 107/375. That night, at mail hour,
it saw a bunch of mail from some other node (375 in net 1)
and marked them all as orphans. Then, since it had no
instructions to forward mail for 1/375, it left them sitting
there. I suppose I could have set my ROUTE.BBS to accept
mail for forwarding from 1/375, but I didn't get around to
trying it. My advice is to just sit tight and wait for your
mail to go out, and then bring 10i up before entering any
new messages.
FIDONEWS -- 06 May 85 00:00:52 Page 3
============================================================
NEWS
============================================================
**********************************************************
Sensible Piracy Protection
**********************************************************
4/21/85 13:00
Just this morning, I was thumbing back through my latest
version of PC Magazine and came across a very unusual ad.
It is put out by Peter Norton. In this ad, he extols his
product, the Norton Utils. Version 3.0, which if you have
ever used these fine utilities will agree with me that
they are a very handy little item to have in your toolbox
of goodies.
These, like many other fine programs of its ilk, are not
copy-protected. So far, so good. We at least have a step
in the right direction. But it wasn't so much the fact
that he changed his ad that caught my eye, but rather, it
was the little blurb at the bottom about upgrades. To wit:
Got an old version? Upgrade to
Version 3 for $25. Got a pirated
copy? Go legitimate. Get the new
version and your own manual for
the upgrade price. Just send a
check and your old disk.
A new twist in the ever-increasing heated battle against
software piracy? YOU BET. I applaud Mr. Norton for his for-
ward-thinking attitude about the problem, and a rational,
thought-out approach to the problem versus that of Mr. Mitch
Kapor who is hellbent on snagging every last pirated copy of
Lotus and Symphony on the market today, and I can assure you
that that number is quite large. Mr. Norton doesn't employ
stupid copy-protection garbage like Lotus. You are free to
make as many copies of the programs as you desire. Instead
of punishing the pirate, he is giving them a chance to "go
legitimate". I might point out that the current version of
the programs goes for the tidy little sum of $99.95. So all
in all, what we have here is a very good approach to the
problem, instead of a knee-jerk reaction the oft-inflated
over-blown problem of software piracy.
Ultimately, the software industry will have to change. Users
with super XT systems packed with megabytes of storage will
no longer tolerate the inexcusable approach of only one
copy on your hard-disk per floppy. There exists only one
form of copyprotection that can not be broken (yet) and
that is through the use of ROM-based software. And if the
software publishers and authors think for two nanoseconds
that I'm gonna sit still for that (and this applies to you,
too, gentle reader) they have another thought coming.
I, for one, will take my business elsewhere. Copy-protection
is ultimately a means of defeating the purpose of software,
and any kid with two bits worth of knowledge on a computer
FIDONEWS -- 06 May 85 00:00:55 Page 4
can crack just about every scheme on the market today.
Granted, there is one drawback to this problem and that
is one relating to the legitimate users who actually pur-
chased the software. I fully understand the objections to
offering pirates the same benefits as a legitimate user,
but nonetheless, I still believe this is a positive step
in the right direction.
Perhaps this is a trend of things to come. I don't know that
for certain, but it certainly is an encouraging sign. Per-
haps the software authors of the world will eventually
learn. That remains to be seen.
Vic Bachulis, Sysop
Washco Fidonet Node 59
**********************************************************
Please direct all flames to /DEV/USR/NUL and all rational
thoughts and comments to Sysop, Fido 59, 1-503-643-2284
**********************************************************
------------------------------------------------------------
FIDONEWS -- 06 May 85 00:00:56 Page 5
Freeware
David Kaplan, Fido #387
The mainstay of bulletin boards is freeware. However, those
of us that use public domain software at work have a problem
many private users are not aware of.
Corporations are extremely sensitive to the source of their
software to ensure that any proprietary programs are
legitimately licensed. If freeware comes with a ".doc" file
indicating this fact, or the application itself displays
such a message there is no problem.
The problems arise when there is no indication that a piece
of software is actually in the public domain. Ideally, the
freeware message and the author's name and address would be
desired to indicate this fact.
This could appear in any of 3 places:
1) Part of a ".DOC" file, or more preferably,
2) Part of the application's screen display, or
3) As a 'display' message that can be seen if the .EXE or
.COM file is typed.
Without this information, many companies are quite hesitant
to use and perhaps enhance public domain software. Many
utilities that we have at home are unavailable to us at
work.
It is an impossible task to track down existing freeware
authors, but if we start following this concept for all new
software, the authors will get the recognition they deserve.
------------------------------------------------------------
FIDONEWS -- 06 May 85 00:00:57 Page 6
========================================================
**** Reply to Fido 464 and all IBM BBS Fido sysops ****
========================================================
In reply to Node 464's request for swapping software via
routing boxes of disks around the country, I'm in full
support of the idea. I currently have around 12 - 13 MB
of IBM-only software collected from around the country
which cost me in the neighborhood of around $800 to $900
to download, and while it does represent a very nice bit
of stuff, I find I'm running out of new stuff at an alar
ming rate. If there are any takers out there who are in-
terested, please let me know. Just about everything I
now have has been tested for basic soundness and useful-
ness and has withstood the test of time by hundreds of
users in my area (Fido 59 is basically the only IBM sys-
tem in this neck of the woods). If any of you wish to
do some swapping, by all means let me know. I always can
use new stuff! There are a lot of top-quality programs
that I have that are begging to be spread around the US.
While I'm on the topic, I need to have Fido 464 contact
me via the net because I can't find a very current node-
list....464, consider yourself paged!
Please respond and direct all inquiries to:
Sysop
Fido 59, Washington County Fidonet, Node 59
Beaverton, ORegon
1-503-643-2284
Thanks much. Vic Bachulis, Sysop.
========================================================
------------------------------------------------------------
FIDONEWS -- 06 May 85 00:00:58 Page 7
MODULA-2
We are looking for Modula-2 programs or users running
Modlua-2 for the purpose of public domain program exchange.
Comments on impressions of present compilers and systems
are welcome at Fido node #418. We are interested in finding
out anything relating to this new language.
We would also like to hear from anyone who might know
any news or have comments on the "Revisions and Amendments
to Modula-2" by N. Wirth published in the "Journal of
Pascal, ADA & Modula-2", 85 Jan/Feb, Vol 4 No 1 issue.
We have Logitech Modula-2/86 source available for M2MDM,
a XMODEM - MODEM7 package for the Sanyo MBC-55x running
under MS-DOS 2.11. It supports up to 4800 bps and is an
example of interrupt processing using Modula-2 co-routines.
It lacks, at present, phone book and dialing features. So
dialing is manual from the keyboard if your modem supports
it. Full CRC error checking and batch transfers compatible
with Fido are included.
Also available is SYMDUMP for examining the contents of
compiled ".sym" files. SymDump is quite portable and is
especially useful if the ".def" source is not available!
Lloyd Miller and Rick Keppler
Calgary_Fido, node #418
------------------------------------------------------------
FIDONEWS -- 06 May 85 00:00:59 Page 8
PC-Rockland, located in scenic South Nyack, New York is one
of the many RBBS systems around. However, few if any can
claim all the features we found when we called into the
system.
The Sysop currently is running 3 nodes simultaneously, with
the capacity for 1 more node immediately available when
needed.
The system currently supports 2400 baud, as well as the
usual 1200 and 300 baud connections. As of last night the
total storage space on the system exceeds 250 megabytes, all
of which is on-line, available storage.
The Sysop has networked an IBM AT using PC-NET along with 3
IBM PC's to form the nucleus of his system. The AT has a 60
megabyte internal hard drive, and an external 'tower' drive
with tape backup. He happens to be a BETA test site for RBBS
so you will ALWAYS find at least the most current version of
RRBS running, and usually an advanced version, not yet
released.
At last count there were over 40 directories of software for
downloading, with literally thousands of programs available,
so every caller can find something he likes there.
If you are interested in calling this unique system it is
running 24 hours a day at (914) 353-2176, and all callers
are welcome to call on that number. Once on the system the
other numbers and associated information is displayed for
you.
------------------------------------------------------------
FIDONEWS -- 06 May 85 00:01:00 Page 9
RUNNING FIDO ON THE SANYO 550
WAYNE CONRAD
FIDO #452
Having just set up FIDO to run on my SANYO 555 with the
Anchor Mark XII Modem, I found it to be quite an
experience... I had problems with the DOS, problems with
the serial interface and problems with the modem. With some
fiddling around and a lot of help from the Sysops of the
other SANYO BBS's around the country, I finally did get FIDO
to work correctly with this configuration. I figure maybe
this article will help anyone else who wants to try it.
What the heck, maybe my instructions will even be correct!
COMMAND-LINE SWITCHES
The first thing I had trouble figuring out was the /V switch
which is used in the command line to tell FIDO which bit of
the serial port's status byte is used as the carrier detect.
With the SANYO it is bit 7, and the mask for bit 7 is 2^7
which is 128, so your /V switch will be like this:
SYO_FIDO 128/V
Of course you will need to include other switches in the
command line for things like download limits and so forth.
We're using bit 7 as the carrier detect because it's the
only status bit presented by the 8251A which controls the
serial port. The 8251A simply accepts this signal and
reflects it's status in that bit, nothing more. If you have
the SANYO serial card, then the signal presented to the
8251A is the DTR (Data Terminal Ready), not the Carrier
Detect. You can tell if you have this problem because FIDO
will either never answer an incoming call, or it will answer
calls that don't exist. If this is your problem, then read
on; I'll present some solutions that should work.
SANYO SERIAL CARD PROBLEMS
In the SANYO serial card, the signal represented by bit 7 is
DSR (data set ready), not CD (carrier detect) as FIDO
expects. This will either cause FIDO to never answer the
phone, or to always answer even when there`s no connection.
What you need to do is to present CD to the serial card
instead of DTR. There are a few ways to do this that I can
think of:
o Modify the modem cable
o Use a Null Modem or other adapter
o Get a serial card that allows you to select CD or DSR
by using jumpers
FIDONEWS -- 06 May 85 00:01:02 Page 10
o Modify the serial card you have
MODIFY THE MODEM CABLE
On the computer end of the modem cable there are two signals
we are concerned with here:
Pin 6 -- DSR (Data Set Ready)
Pin 8 -- CD (Carrier Detect)
Oh by the way, you need to have your modem set up to present
a "true carrier" on the CD Line. I don't know how this
works, since the Anchor always presents a "true carrier,"
but I guess that with your modem set for "true carrier," the
CD line goes active only after the two modems have connected
and have good carriers on the line. Probably with a "false
carrier" (?) the CD goes active when the phone rings or
something like that. In any case, some modems have a switch
for this; make sure it is set for "true carrier."
What we want to do is to present the CD line to pin 6 rather
than the DSR line. If your modem cable has a solder-on end,
then you can probably do the job with a soldering iron in
just a few minutes. Just Unsolder the DSR line from pin 6
of the connecter (remember this is the connecter on the
COMPUTER END of the cable). Put a little electricians tape
or silicon jel on the now freed DSR line to prevent it
causing a short. Unsolder the CD line from pin 8 and
resolder it to pin 6. Be careful of course to check for
solder bridges and the like; Modems (and computers) get
quite upset about shorts and may punish you with DOWNTIME
and REPAIR BILLS.
If your cable is ribbon cable with the press on ends, you
will need to remove the end. This is a bit tricky, but it
can be done without destroying the connector, if you're VERY
VERY careful and if the connector cooperates with you. You
need to use a screwdriver or something to lift up the
locking parts of the connecter until it comes apart. You
might want to have a spare handy just in case; I've
destroyed more than one cable end this way. Before you
remove the ends, carefully identify which line goes to pin 6
and which line goes to pin 8 and mark them with a felt tip
pen or something. THEN remove the connector. Now you need
to split the cable to separate lines 6 and 8 from the
others. Take line 6 and 8 and exchange their places. You
might use some tape or something to keep all this stuff in
place (your cable is now split into about 5 different
parts). Then carefully align the cable on the disassembled
connector, put the connector top on, and press them
together. There is a big expensive press used for this, but
you can probably make do with a vise and a few small pieces
of wood or whatever else you can think up of. Make sure you
get the connector on the right way. Get it on the wrong
side of the cable or turned 180 degrees and nothing will
FIDONEWS -- 06 May 85 00:01:05 Page 11
work.
Actually, this is such a pain, and so prone to error, that I
probably wouldn't attempt it unless there was absolutely NO
other option available.
NULL MODEMS and OTHER ADAPTERS
You can make an adapter by getting one female and one male
DB-25 connector and some wire. You can do it with ribbon
cable by using press-on connectors, pressing the male
connector on one end, then splitting the cable on the other
end and doing the funny stuff above to it before pressing on
the female end. Again, it's probably better to use solder-
on connectors and shielded cable.
Connect these pins straight-across:
7 Signal Ground
2 Transmit
3 Receive
20 Data Terminal Ready
4 Request To Send
5 Clear to Send
Then you will connect pin 8 on the male (modem end)
connector to pin 6 of the female (computer end) connector.
Also, make sure the shield is connected to both of the
connector cases. Pin 1 of the rs-232 "standard" is used
for protective ground; however I'm not sure whether this
should be connected with the shield or not. It's been a
while since I've make RS-232 cables. And I hope a long
while more.
There are also these nice (and usually expensive) little
boxes called NULL MODEMS, which are basically a switchbox
with a male connector on one end and a female connector on
the other, so you can configure it any way you like. If you
are lucky enough to have one of these, set it up like I just
described above.
GET A BETTER SERIAL CARD
Not a bad idea at all. The better serial cards have jumpers
on them which allow you to decide which lines are used. All
you have to do to get the serial card to recognize CD
instead of DSR is move a jumper or two... We should all be
so lucky!
MODIFY THE SANYO SERIAL CARD
This is what I chose to do. The modem cable is build into
my Anchor, and I didn't want do tear up that ribbon cable
anyhow. And I wanted it done NOW, not after going to the
FIDONEWS -- 06 May 85 00:01:07 Page 12
store for parts to make an adapter and so on. This
modification takes an exacto knife, a soldering iron,
solder, and a small piece (two inches or so) of wire--I
prefer wire-wrap wire.
Take your serial card out of the computer. Now look at the
end of the card where the connecter is soldered to it. On
the connecter all the holes that the pins fit into are
numbered. Look for number 6 and number 8. Follow these
onto the card and see where they connect to it. On the top
of the board you will see a trace coming from pin 6. Cut
this trace with the knife. Now take a look at where the
trace goes to. Holding the card with the connecter away
from you and the components on top, the trace goes to the
NEAR-RIGHT pin (pin 1) of the IC at the FAR-LEFT of the
card. Turn the board over and solder one end of the wire to
that pin. Solder the other end to the pad where pin-8 of
the connector comes onto the card and you`re done. This
might not seem too clear, but just be careful with following
the pins onto the board and so on so you get the right ones.
DISCLAIMER
Please note that I wrote these instructions AFTER THE FACT,
so I cannot guarantee their accuracy. Also, as far as I
know all SANYO serial cards are the same, but there may be
look-alike-but-cost-less cards out there that have their
board laid out differently. So please be aware.
THE DTR PATCH
You will also need a patch for your DOS. both MS-DOS and
DS-DOS fool around with the DTR line during disk accesses,
this is a no-no since it causes many modems to hang up.
There is a patch you can get to fix this. There is one
patch for MS-DOS 2.11 and another patch for DS-DOS 2.11.
This patch only needs to be run once each time you boot, so
you can put it in your AUTOEXEC file. it goes by various
names but the ones I've seen are DTRPATCH.COM and
SYOPATCH.COM.
You can get the patch from FIDO #1 or many of the SANYO
bulletin boards across the country. If you can't find it
anywhere else let me know and I'll send it to the Nearest
Fido in Your Area via netmail. It's pretty small anyhow; I
think 3k for both the MS-DOS and the DS-DOS version with a
small doc' file.
BATCH FILES FOR FIDO
This is how I set up my RUNBBS.BAT file:
:LOOP
FIDO_SYO 128/V
FIDONEWS -- 06 May 85 00:01:09 Page 13
IF ERRORLEVEL 1 GOTO EXIT
GOTO LOOP
:EXIT
And the AUTOEXEC.BAT file is something like this:
DATE
TIME
DTRPATCH
I also have a LOCAL.BAT which I use to bring up fido in the
local mode:
FIDO_SYO /T
THE ANCHOR MODEM'S DTR PROBLEM
Most modems use DTR to hang up the phone. The Anchor Mark
XII ignores DTR. This is a problem because FIDO expects the
modem to hang up when DTR is dropped. Now, there is a fix
(requiring a transistor, a resister, and a few minutes) to
cause the Anchor to use DTR normally. I haven't done it
yet, but I might.
What happens when FIDO is sending netmail and files is that
it tries to hang up the phone when it is done by using DTR.
If it can't, or if it even suspects that it can't, it will
start beeping like crazy for you to wake up at 2am and fix
it. What a pain, but it does save you from horrid phone
bills should something happen to prevent a proper
disconnect. What I've noticed is that sometimes the
disconnect will occur but FIDO will wake me up anyhow, while
other times the disconnect will occur just fine and FIDO
won't fuss at all. I don't know why this happens, but I'm
going to modify my modem anyhow to see if it helps.
Actually, what I suspect is happening is that FIDO toggles
DTR to hang up, waits a few seconds, then checks to see if
the carrier is still there. If the remote FIDO has hung up
by then (which it usually has) then there's no carrier and
no problem, otherwise FIDO does it's beeping thing... Of
course while it's beeping away the other computer has hung
up or the smart commands FIDO issues as a backup have done
their thing, but FIDO is faithful and wants to protect your
phone bill so it wakes you up to make sure.
You can get the correct instructions for this fix from Jon
Tara of FIDO #92. He also published these instructions in
FidoNews 203, but please be aware that the instructions as
published in the FidoNews are INCORRECT. The corrections
were published in FidoNews 315, but it's probably easiest to
get the corrected instructions from #92 directly. By the
way if you get the corrected instructions from #92, I
wouldn't mind having a copy (hint, hint)...
FIDONEWS -- 06 May 85 00:01:12 Page 14
CTTY PROBLEMS (STILL)
Here's one I still haven't figured out; the CTTY command
will not work properly. What I did to test it is to connect
with a friend by using the MINITEL communications program,
then I dropped into DOS without losing the carrier and
typed
CTTY AUX:
which went over like a led zeppelin. Didn't accept anything
from my friend or give him anything back. I don't know if
all my patches and mod's had anything to do with it, because
I had never tried it before I went in with the knife and
soldering iron.
I need for this command to work if I'm going to implement
FIDO's 0 command to allow the DOS to be run remotely. Also,
I haven't even tried Watchdog to see if it runs on the SANYO
but I really doubt it will. The Serial port is one of the
SANYO's poorest compatibility hangups.
What I might do if nothing else works is to write a small
program to take the place of the CTTY redirection and the
watchdog. Here's my ideas on that so far:
o Be invoked from the batch file RUNBBS.BAT.
o Redirect standard console input and output to it's own
device drivers. The output driver would write to
both the local screen and to the remote system. The
input driver would read from either the keyboard or
from the remote system. This would allow
simultaneous typing. Also you could see what the
remote Sysop is doing to your poor system...
o Load in a copy of COMMAND.COM and run it using the
redirected i/o drivers.
o To get back into FIDO, just type EXIT which would
return control from COMMAND.COM to the supervisor
program, which would then restore the standard output
drivers and drop back into the RUNBBS.BAT.
o The Supervisor program would check the carrier
whenever something is output. If the carrier is lost,
then it just aborts COMMAND.COM and drops into the
RUNBBS.BAT. This avoids the rather drastic re-boot
that Watchdog does when it loses the carrier. It is
important to those of us without hardware clocks to
avoid re-booting.
Of course this is not a two-hour project, especially for
someone who's never redirected standard output or run .EXE
files from withint a program. What I'm saying is, if anyone
wants to assist in this endeavor I certainly won't refuse
the help. This would probably work for the IBM without
FIDONEWS -- 06 May 85 00:01:14 Page 15
modification, so you IBM hackers keep that in mind; I can
use help from ANYBODY experienced in assembly language with
the things I mentioned above.
FOR FURTHER INFORMATION
RS-232c Specifications:
Elizabeth A. Nichols, Joseph C. Nichols and Keith R. Musson,
DATA COMMUNICATIONS FOR MICROCOMPUTERS (McGraw-Hill, 1982)
Schematic for the SANYO Serial card:
Robert Kong Win Chang, "Build a Serial Card," Byte, vol. 10
no. 3 (March, 1984), 129
Modifying the Anchor Mark XII Modem:
Fido Newsletters 208 and 315, or
Jon Tara of FIDO #92
------------------------------------------------------------
FIDONEWS -- 06 May 85 00:01:15 Page 16
From: Mark Welch on Fido 459 (WelchNet, San Francisco, CA)
A Telecommunications Dictionary
-------------------------------
Tom Jennings at one point brainstormed the idea of a
"telecommunications dictionary," essentially using the known
frequent content of BBS information to reduce the time it
takes to transfer text from a BBS to the user's system.
The basic idea is that frequently-used words or
substrings (words like "message" or "the") could be encoded
in two bytes instead of sending the full strings. If the
'dictionary' were flexible enough, it might include an
entire screenful of information encoded as two bytes (a
standard menu or help screen, for example, or a log-on
logo).
I toyed with the ideas, and juggled with some of them a
bit, and wrote a quick and dirty program to test the basic
theory. While Tom had the idea of a "learning dictionary"
in which the user's software could learn new encoded strings
from the host as necessary (and in which each host would use
a different dictionary), I tended toward the idea of a
"standard" string dictionary.
By way of background: standard "squeeze" programs use
something called Huffman coding to encode strings by using
fewer bits to encode characters which occur more frequently.
Huffman coding doesn't work with interactive text because
(a) it works best on long, arbitrary text files, and (b)
Huffman coding abandons the meaning of byte boundaries, so
that one dropped bit in a transmission means the entire file
is lost unless some sort of error detection and correction
(ECC) is included. ECC eats up as much or more bits than are
saved.
My approach was to examine text files containing captured
message bases from a Fido BBS and several other on-line
systems. I analyzed the text [using a program quickly
written in Turbo Pascal] and found 32 strings (4-, 5- and 6-
bytes long) which occurred very frequently. A
telecommunications "dictionary set" might include 256 words
in each of several dictionaries; in theory, system-specific
dictionaries might exist (Fido, CompuServe, etc.).
My dictionary program replaced strings like " the "
(space,t,h,e,space) and "ing " (i,n,g,space) as two-byte
codes. Rather than write a complete BBS and telecomm
software to test the theory, my program merely encoded the
information from one file into another file. A real
telecommunications dictionary would have to do this encoding
(at the host end) and decoding (on the user's end) "on the
fly," adding a number of potential problems if existing
telecomm software is running at or near the capacity of the
CPU.
FIDONEWS -- 06 May 85 00:01:18 Page 17
My dictionary program, using 32 strings of length 4-, 5-,
and 6-, re-coded those strings as two-byte codes: a prefix
byte (with a value greater than 128) and a suffix byte from
0..31. A "full" telecommunications dictionary would do the
same but might work from as many as 127 dictionaries (using
the full range of the prefix byte) and as many as 256 words
in each dictionary, or over 32,000 possible strings.
The result: By encoding the 32 most-frequent strings, I
reduced the file size -- which would translate to
transmission time -- by almost exactly ten percent. That 10%
savings was constant over about a half-dozen files, ranging
from Fido messages to a news column I was writing.
I would estimate that a "generalized" dictionary of 256
words might save around 15%. To obtain any significant
savings from the encoding, a system-specific dictionary
(i.e. Tom's "learning" dictionary, or a Fido-specific
dictionary) or a very large standard dictionary would have
to be used. [An important aspect of this is that use of a
single, standard 128-word dictionary by encoding into one
byte rather than two would save about 25%, but that savings
could never be enhanced later.]
Savings are always enhanced by encoding long strings
instead of short strings (i.e. if "message" occurs often and
is encoded, you save more bytes than by encoding "the").
A note: software which used some sort of dictionary-
coding to save on transmission time would have to "register"
its capability during or after sign-on, so that BBSs or
services with the dictionary available could still default
to a "dumb terminal" mode. If the software used a "learning
dictionary," it would have to be able to signal "I don't
know that word" at any time; the host would then download
the word to be added to the system's dictionary. Most
essential: all of this ought to be automatic, transparent to
the user.
I'd like to know if anyone else is interested in
following up on this idea of a "telecommunications
dictionary," and what people's ideas are on this. This was
Tom's brainstorm but he had other projects and has put it on
the shelf; likewise, I'm not actively pursuing it further
right now, but thought I should contribute my experience to
other telecommunicators. Please let me know if there's
interest in this topic; send Fido mail to me at Fido #459
(415-391-0556 6pm-9am PST only), or write to me c/o 425
Battery St., 4th Floor, San Francisco, CA 94111.
-- Mark Welch, Fido459
FIDONEWS -- 06 May 85 00:01:20 Page 18
============================================================
COLUMNS
============================================================
From: Tim Meyers, Fido #365
Dated April 25,1985.
Tim Meyers
Techn-Source Systems Group.
Dear Mr. Meyers:
Thank you for your interest in becoming a service
provider for Compuserve. The first step to becoming an
information provider or a system operator for a Special
Interest Group is the preparation of a proposal
describing the proposed service.
Taken from a recent letter I received from Compuserve
Information Services, a proposal to start a Special Intrest
Group for Fido. It will be labeled as (GO FID-1) when it
arrives, but first I have to know exactly how many people
would be interested.
My company will invest $5,000 for the initial fee for the
creation of the database and files. After which point it
becomes the property of me, and or whoever wants it.
I have typed a proposal to Compuserve and should get a
response within 60 days. I spent about 20 hours on the work
of menus, methods, promotional plans, target market, current
projection, overview of information etc. etc. Now it is
completed and I hope that I haven't wasted my time. Here is
a brief review of what will be on the SIG:
1st, the newsletter. At this point in time several
people are just not receiving the newsletter. It's very
large to download and many people just cannot access the
larger systems that have it online. The newsletter
would linked to a electronic magazine, in which menus
serve as the "Table of Contents" and point to articles
that have been written. It works very well for the
needs of both users and sysops.
2nd. Mr. Jennings would create a nice little "Welcome
to the Fido Special Interest Group". Just a simple
letter explaining what Fido is, what's so unique about
it, etc.
3rd. Users Questions Answered. That's it, the biggest
problem at this point in the network. It would be a
great place for all questions and answer to be placed at
Compuserve's cost. Everyone has asked for it, now anyone
that wishes to chip in can. Questions answered in a
day, No more 100 calls to Tom Jennings. Keeps small
questions like "What do I do with this modem?" from
costing Mr. Jennings 25 cents. Etc. Etc. Etc. I am
sure you can see the need for it. At this point mostly
FIDONEWS -- 06 May 85 00:01:24 Page 19
all new sysops questions are addressed to Mr. Jennings.
This way anyone that has an answer can solve somone's
problem. Not just sysops but from a user's standpoint
also.
4th. Software. Every single software program related to
Fido will be placed on the system. A message will be
placed for which version is the newest, when updated
etc. I made three phone calls tonight, one to node #1,
got the newsletter, one to #375 for this article, one to
51 for the TWIX program. When I could have made just
one call to Compuserve, and it's a local charge.
This could solve lots of problems with disk space on
most systems, although I am sure many sysops will
continue to keep Fido software online, they can at least
have a common conference area. It's a place to keep in
touch.
5th. The Nodelist, Each week it could be put online.
Instead of spending the many hours of re-dialing and
busy signals you would be able to obtain it whenever you
wished. I am sure the sysops of fido #50 would see the
need of this.
6th. What does it cost? Well there is a one time
connection fee and several other addition cost, but I am
signing the check here. I will sign a check for
$5,000.00 to form the Fido SIG if I receive enough
interest in the project. I wish to help the many other
sysops and future sysops of Fido, including the users.
We need somthing like this. The problem is, how many
people would be interested? With over 250 nodes and
growing, again it's needed.
My Plan of attack? What is my attitude? Well here are a few
simple figures:
Fido SIG operates at 10% royalty rate. If Each node had a
cumulative of 60 minutes in the SIG it would have approx.
253 hours of use. This is resonable since many SIGs draw
less than 100 hours a month, which is a useful amount. At
253 hours a month, Compuserve will be happy to let Fido
remain on the system But if it falls to less than 25 hours,
I start paying $750.00 a month. That's a chance I will have
to take, but first I have to consider how many people use
the system.
1) Each node -- users, sysop, etc. -- spend 2 minutes a day
on the Fido SIG and we have our goal.
2) Half the nodes spend 2 minutes and we have 2/3 our goal.
3) 50 nodes spend 2 minutes on the SIG and I have to spend
about $115.00. Not bad, At least we have provided as
much information and equipment that we can to make Fido a
FIDONEWS -- 06 May 85 00:01:27 Page 20
nation wide project.
All users and Sysops, Please Send any Ideas, Comments, or
criticism to The Sysop Fido #365 or #442. Thanks Tim Meyers.
FIDONEWS -- 06 May 85 00:01:27 Page 21
============================================================
FOR SALE
============================================================
USRobotics Courier 2400
"The Intelligent Choice in Data Communications"
__________________________________________
/ |o o o o o o o o o| COURIER \
/ |_________________________| 2 4 0 0 \
/ HS AA CD OH RD SD TR MR AL | | || |||| \
/________________________________________________\
\______________________________________/
"So delighted am I that I'm ready to trade in my
1200 baud Hayes Smartmodem."
-Henry Kisor, "The Chicago Sun Times"
"The Courier 2400 performed so transparently and so
steadily that it was almost immediately adopted as
the remote unit used to test the other modems..."
-Emmitt Dove, "PC Products Magazine"
"The new 2400 baud Courier modem is consistently
dependable, and has many thoughtful features which
no other modems have."
-Peter McWilliams, "The McWilliams Letter"
"USRobotics' 2400 baud modem doubles throughput and
enhances the Hayes standard at a terrific price."
-Jeff Duntemann, "PC-Tech Journal"
(product of the month, May'85 issue)
The Price, Just $699 list.
Contact: USRobotics, Inc.
1123 West Washington
Chicago, IL 60607
(312) 733-0497
FIDONEWS -- 06 May 85 00:01:28 Page 22
============================================================
NOTICES
============================================================
*** Calendar of Events ***
12 May 85 Submissions deadline for next issue of Fidonews.
17 May 85 The Next Occasional NY/NJ Sysop Meeting.
27 May 85 through 31 May 85; Spring 1985 DECUS symposium,
New Orleans, LA. Among other events, Kurt Reisler
(sysop Fido 74) will give a 1 hour talk on Fido.
If you have any event you want listed in this calendar,
please send a note to node 107/375.
Download original FidoNews · Volume 2 (1985) · ← Previous · Next →