textfiles/bbs/CBBS/1991/910915.ch

CBBS(R) 4.0.3b
09/15/91 04:35:19

Y/N: want CBBS "1st time user" info?^U
?^U
?^U
?n;ward;christensen;odraw;;fullc;piss

Logging name to disk...
You are caller 223717; next msg =45148;  387 active msgs.

NO CARRIER
pizzaCBBS(R) 4.0.3b
09/15/91 04:41:29

Y/N: want CBBS "1st time user" info?^U
?^U
?^U
?n;ward;christensen;odraw;;fullc;piss

Logging name to disk...
You are caller 223720; next msg =45148;  387 active msgs.
Prev. call 09/15/91 @ 04:40, next msg was 45148
Recording logon for next time...
Use FULL? to check assignments
?^U
?xxxxx
"Mine" command checking for msgs TO you,  ^K
>Function:?dir c:log;dir c:killed;dir summary;type-20 log,ward c;or;*;short
LOG.         13 
KILLED.      17 
SUMMARY.     25 
09/15/91,04:35:23,223717,2,WARD CHRISTENSEN,,
09/15/91,04:36:11,223718,2,WARD CHRISTENSEN,,
09/15/91,04:40:39,223719,2,WARD CHRISTEN
#
Next caller 223722; next msg =45148; 387 active msgs.
*=45148 (hi msg last call)
>Function:?or;45131^U
?or;45131
45131 09/11/91 WARD CHRISTENSEN => JERRY OLSEN: "R/REMOTE COMPUTING"
45132 09/11/91 WARD CHRISTENSEN => ALL: "LZH?"
45133 09/11/91 ERIC BOHLMAN => WARD CHRISTENSEN: "R/LZH?"
45134 09/12/91 WARD CHRISTENSEN => ERIC BOHLMAN: "R/LZH?"
45135 09/12/91 AL HIGGINS => LANE LARRISON: "R/DIABLO 620 INTERFACE"
45136 09/12/91 AL HIGGINS => RICHARD HINTON: "R/OS AND BEYOND"
45137 09/12/91 MAX KOHN => ARNOLD BOYD: "XT COMPUTER"
45138 09/13/91 BILL MATTSON => RICHARD HINTON: "R/OS AND BEYOND"
45139 09/13/91 BILL MATTSON => RICHARD HINTON: "APPEND"
45140 09/13/91 TOM ULBRICHT => ALL: "FILECOMMAND II FOR DOS 5.0"
45141 09/13/91 CARLTON MOORE => WARD CHRISTENSEN: "R/CNE?"
45142 09/13/91 JERRY OLSEN => MARTIN SCHRADER: "R/REMOTE COMPUTING"
45143 09/13/91 JERRY OLSEN => ROY LIPSCOMB: "R/VOLUNTEERS NEEDED"
45144 09/13/91 JERRY OLSEN => WARD: "R/REMOTE/ LZW"
45145 09/14/91 CHARLIE KESTNER => PETE GROBAREK: "R/YOUR UNIX BOARD"
45146 09/14/91 DAVID GIBBS => ALL: "AS400 POSITION WANTED"
45147 09/15/91 ERIC BOHLMAN => WARD: "COMPRESSION"

 - End of summary - 

Retrieving flagged msgs: C skips, K aborts.

Msg 45131 is 10 line(s) on 09/11/91 from WARD CHRISTENSEN
to JERRY OLSEN re: R/REMOTE COMPUTING

 What is your remote application? When I first saw REMOTE, in about
maybe '86, I dialed into another guy's system, using ADM-3A emulation!
 I am looking at my PC Anywhere book, and sorry, it is from '86, so
things may have changed, but they HYPE the point that you don't HAVE
to even have a PC to call in.
 Of course if you need graphics, and if the 25th line "hokeyness"
necessitated by calling in with a 24-line screen are a problem, or if
you don't like having to press "esc-1" for F1, or "esc-U" for up-arrow,
then this isn't for you. Depends upon the application, frequency of use,
need for convenience, etc.

Msg 45132 is 30 line(s) on 09/11/91 from WARD CHRISTENSEN
to ALL re: LZH?

 Has anyone else "putzed" around with LZH algorithms? I started with
a paper on it, and took the pseudo-code example it had, and implemented
it in my favorite programming language - PMATE editor macros (took under
300 keystrokes to code a full LZH encoding algorithm - this is a VERY
net programming language!).
 The output is basically a bunch of numbers. The decoding program starts
with a table consisting of the possible codes in the "set" (such as
ascii, or all 256 possible binary values) and starts decoding.
 The thing I was wondering: do any lzh routines do a 2-pass technique,
in which, after encoding the LZH, they then Huffman encode the resulting
NUMBERS?
 Also in Pmate, I wrote an output number frequency histogram, and it seems
quite "clumpy", implying Huffman encoding would do a lot to further reduce
the output data size.
 Of course, other tricks could be used: if the output is a byte with
the 8th bit off, it is "straight ASCII", while if the 8th bit is on, then
it is perhaps a 11- or 15-bit value (ignoring the 8th bit) which represents
a point in the code table. Here's the Pseudo-Code.
[1] Initialize string table; (to 256 entries, 00-FF for example)
[2] [.c.] <- empty;
[3] K <- next character in charstream;
[4] Is [.c.]K in string table?
 (yes: [.c.] <- [.c.]K;
 go to [3];
 )
 (no: add [.c.]K to the string table;
 output the code for [.c.] to the codestream;
 [.c.] <- K;
 go to [3];
 )

Msg 45133 is 09 line(s) on 09/11/91 from ERIC BOHLMAN
to WARD CHRISTENSEN re: R/LZH?

What you're describing sounds more like LZW coding rather than LZH
(LZW was the coding scheme originally used in ARC. It's fallen a bit
out of favor because Unisys has a patent on it and recently started
enforcing it). The implementations of LZH that I've seen rely on a
"sliding dictionary" which keeps track of the last 8K or so of input.
If a string of text can be found in the dictionary, it's replaced with
a code consisting of an offset, length pair ("insert the 20 characters
that you saw 550 bytes ago"). These codes are then subjected to Huffman
compression.

Msg 45134 is 14 line(s) on 09/12/91 from WARD CHRISTENSEN
to ERIC BOHLMAN re: R/LZH?

 Rats, yes, LZW. I typo'd.
 Ah, offset, length, 8K dictionary - makes sense.
 I wonder what some sort of "comprehensive multi-look at a file" would
result in - i.e. if you have a sufficiently large file, might it just
strip it into "words", and have a huffman-encoded "word number" or
something like that?
 Lets see, if you have offset/length, that implies the data is strictly
a "string" so to speak - no delimiters. Hmmm, interesting!
 Do you have any idea how the table is intialized? LZW looked
interesting in that you didn't need a decoding table!
 I wonder what MNP does - some sort of on-the-fly compression. I wonder
if it can say pass things thru "raw", then from the first block, construct
an optimization table ON BOTH ENDS, then encode the 2nd "chunk" from the
table, then build-up the table, etc!

Msg 45135 is 05 line(s) on 09/12/91 from AL HIGGINS
to LANE LARRISON re: R/DIABLO 620 INTERFACE

Its been a few years since I've had to work with serial devices. As
far as the 620 goes its a straight through cable, reverse 2&3 and strap
5-6-20 and everything works fine.
The serial card had an error on one of the pins assignments that thru
me a loop for a few hours. Thanks for the help!

Msg 45136 is 10 line(s) on 09/12/91 from AL HIGGINS
to RICHARD HINTON re: R/OS AND BEYOND

We are using an OS/2 system with PM to support a distributive reports
procesing system. If your thinking on moving into OS/2 thinks lots of
memory! We're using 10 Meg at present and may end up moving to 12 meg.
The problem I'm running into is there are very few diagnostic tools available
for OS/2. For example, I need to have modem access into the OS/2 system
so I can see what our customers are doing. However, no one makes a
remote modem access for os/2 with presentation mgr. I've been looking
for six months and haven't found any. One company has asked us to
beta trial their product. However, I was to have a copy two months
ago and do not know when I will see it!

Msg 45137 is 05 line(s) on 09/12/91 from MAX KOHN
to ARNOLD BOYD re: XT COMPUTER

Is that xt computer you showed me when I came with David Johnson to
pick up his T.I. still available? If it is, please let me know as I may now be
interested as wel as ready to purchase it. Thanks.

Max Kohn

Msg 45138 is 16 line(s) on 09/13/91 from BILL MATTSON
to RICHARD HINTON re: R/OS AND BEYOND

Impressions are still in the formative stage, but I can tell you
that it seems to work just fine with MSDOS 5.0 and the dual boot
feature. I'm using QEMM 5.11 with DOS (NOT MS's memory stuff), and
get 613K on the DOS side. This was 621K before, but OS/2 requires
an APPEND statement in the DOS CONFIG.SYS file to support the dual
boot feature. Note that DOS emulation in the "compatibility box"
only provides 512K, and IBM advises that this is all you will get
in emulation mode. Maybe dual booting will stay around long after
OS/2 ascends to the lofty position of "operating system of choice"
.any bets on when that might be??? Ihaven't installed HPFS yet,
due to inability to get two (old and new) IDE disk drives to co-
exist. BUYER BEWARE - The IDE (AT) interface ain't a standard.
Various drive makers implement it differently. My IMPRIMIS drive
(now owned by Seagate) won't let the Seagate ST1144A I bought be
recognised (or vice versa, who's to say). IBM support (free for
90 days) has been great so far. More later.

Msg 45139 is 04 line(s) on 09/13/91 from BILL MATTSON
to RICHARD HINTON re: APPEND

Sorry, I made a mistake in the previous message. The APPEND statement
needed to support dual boot is in the DOS AUTOEXEC.BAT file. not
the CONFIG.SYS. To be honest, I haven't tried to LOADHI the APPEND
yet, but should. I'll clarify this real soon, now.bye.

Msg 45140 is 08 line(s) on 09/13/91 from TOM ULBRICHT
to ALL re: FILECOMMAND II FOR DOS 5.0

I am trying to get a copy of File Command that works with DOS 5.0.
I have talked to the author (an IBMer) who says it has been upgraded
to work with 5.0 and is available within IBM yet the original distribution
channel (Personally Developed Software) is refusing to support it as is
the IBM software support group. Apparently it uses some BIOS calls or
undocumented system calls that are no longer valid. File Command makes
you're PC files look somewhat like a VM mainframe system.
 Thanks in Advance - TLU

Msg 45141 is 02 line(s) on 09/13/91 from CARLTON MOORE
to WARD CHRISTENSEN re: R/CNE?

Where are you taking your tests and have you taken any
of the training courses?

Msg 45142 is 04 line(s) on 09/13/91 from JERRY OLSEN
to MARTIN SCHRADER re: R/REMOTE COMPUTING

Thanks for your reply. But as luck would have it, I stumbled across a
shareware product called TR3-9 which appears to meet my needs rather
well.BTW, in the past one or more of the commercial alternatives
offered two-machine licenses. But I couldn't recall which.

Msg 45143 is 04 line(s) on 09/13/91 from JERRY OLSEN
to ROY LIPSCOMB re: R/VOLUNTEERS NEEDED

I noticed in your msg here and on my board that the address omitted the
STREET line. Though I have some misgivings about ITRC (which I prefer to
avoid detailing here), my recollection is that their "new" address is at
59 E. Van Buren.

Msg 45144 is 29 line(s) on 09/13/91 from JERRY OLSEN
to WARD re: R/REMOTE/ LZW

My need is to set up a call-in procedure to work clients through their
problems without traveling to them. I've recently been setting up a number
of relatively novice users with new systems. Being able to install their
new software remotely would help.As noted in a prior msg, I've found
a shareware solution which appears to work well. A real asset is that it
requires only a single $25 registration which specifically entitles a
consultant to provide host-machine copies to as many clients as possible.
Great deal, huh?! Only problem I've seen so far (noted in the docs) is
inability to properly function when the host is in graphics mode. But I
don't really anticipate needing that.

The June '90 Dr. Dobbs has code for a modified LZH algorithm in C. I know
I filed the issue around here somewhere, and *THINK* I saved source. I
dabbled with it awhile as a possible basis for compression in a product
I have yet to release. But at least the version(s??) in Dr. Dobbs had
too much of a time penalty for my needs, esp. as compression would have
provided a "bonus" to my program rather than fill an integral need.If
you have need for either the article or source, lemme know. If so and you
would like it sooner rather than later, call voice to 312.939.3300,
leaving a msg if I'm out. I'm receiving a new 'puter tomorrow and setting
it up the hard way-restructuring my entire approach to HD management
from my current machine and thus having to reinstall nearly everything,
incl. the .BAT files I use to invoke virtually every major app. <argh!>
At the same time, I've been swamped with literally about 1K of <printed
pages> from CIS lately. The word is now spreading (too) wildly that I
now have separate columns in the leading shareware pub here in the U.S.
and the leading such pub in the U.K., plus probably will be doing the
same for an Aussie pub debuting in Dec. Everyone wants to get into the act
for free publicity. <G>

Msg 45145 is 13 line(s) on 09/14/91 from CHARLIE KESTNER
to PETE GROBAREK re: R/YOUR UNIX BOARD

 You'll have to go through the newuser program again on chinet.
 Guest callers (and expired contributors) who don't call in a
while automagically have their logins terminated. This also
happens if/when Randy has to restore the system, and he is forced
to use a backup tape that exists BEFORE a caller initially logs in.
(Rare.)
 Contributor dues are $75 per/annum, but aren't necessary to gain
access to off-site email. (However, I'm SURE Randy wouldn't mind
one bit if you sent in twenty-five out of the goodness of your
heart! 8-] ) Contributor status only gains you 24-hour read/write
access to Usenet, and a uucp account if you so wish.
 Guests have access to everything ELSE, along with Usenet reading
privileges (Usenet ONLY on weekends for guests).

Msg 45146 is 13 line(s) on 09/14/91 from DAVID GIBBS
to ALL re: AS400 POSITION WANTED

I am an experienced S/36 programmer with some AS/400 background.
I'm looking for a position in a AS/400 shop.

I am familiar with RPG/400 (RPG III), RPG II, some CL, some DDS,
SEU & SDA on both 400 & 36, plus PC Support on both systems.

Relocation to the right area is a possibility.

If you have or know of a position similar to what I am seeking,
please reply on this BBS or call me voice at 708-437-4717 or 708-635-7100

David Gibbs


Msg 45147 is 26 line(s) on 09/15/91 from ERIC BOHLMAN
to WARD re: COMPRESSION

The "word number" approach would probably give very tight compression for
English text (especially given the Zipf's-law distribution where 20% of
the words account for 80% of the usage). The main disadvantages are
that 1) it would require two passes over the text and 2) the table
of words would have to be transmitted along with the compressed text.
These could be overcome if the compression used a fixed generic word
table, but then the compression would become less tight. Also, it
wouldn't work for non-text material. Classic case of "the higher you
fly, the more towns you can see but the longer it takes to land at
any of them."

The sliding-dictionary implementations I've seen use a variation on
binary search trees which I don't fully understand yet. As characters
come in, you keep them in a circular buffer and also keep a pointer
indicating the last point where you outputted compressed data. You
keep accumulating characters until you come to one such that the characters
from the pointer up to the last character formed a string that could
be found in the buffer, but adding the new one makes something different.
You then output the offset and length for the matching string and move
the pointer up to the new character. Characters that don't match anything
get output literally. While all this is going on, the dictionary tree
is being updated. Decompression is quite simple. Literal characters
just get copied straight through; length-offset pairs result in previous
text being copied forward (you normally buffer the last 8K or so of
text so you don't have to do a file seek and read in order to pick
up the text).
>Function:?
>Function:?