Unix Technical Discussions V1 #9

[unix at cbosgd.UUCP]

From: Ron Heiby (The Moderator) <unix-request at cbosgd.UUCP>


Unix Technical Discussions     Sun, 10 Mar 1985    Volume 1 : Issue 9

Today's topics:
		 Looking for Compupro 68K Driver Help
		Where does Unix use the 8th bit of chars?
		   Need SW to control Logic Analyzer
			C <-> f77 Interfacing
		Looking for LZ Compression Implementations
----------------------------------------------------------------------

Date: 14 Feb 85 21:33:59 GMT
From: stuey at wlcrjs.UUCP (Stu Heiss)
Subject: help needed with 68k

Here's a stickey one.  I've just brought up unix on Compupro
hardware with a 68k cpu using the 68451 mm.  This system is not
currently configurable and I need to be able to do port i/o to a
foriegn board in the system.  The cpu card maps memory requests
to the top 64k to the i/o ports - so you have to be able to
address absolute memory - which the memory management forbids.
Got any ideas? What about /dev/mem or /dev/kmem? Also, I'd like
to get a driver for my Sugart 4008's ready in anticipation of
the reconfig files. I've got low level code written in c, but
not in device driver form. I've no idea what a driver's supposed
to look like - no doc on how to write one. Where do you get this
kind of information? Is there any such thing as standard
skeleton driver code to use as a starting point?
Pleas reply with mail to ===> ihnp4!wlcrjs!stuey	- Stu 

------------------------------

Date: 9 Feb 85 22:17:12 GMT
From: gll at sfmag.UUCP (G.L.Lindgren)
Subject: Use of 8th bit in UNIX System

I am in the process writing requirements for the "internationalization"
of the UNIX System.  One of the key ingredients in this effort is support
for full 8-bit byte streams throughout the UNIX System.  As many of you
know, there are many areas in UNIX that take advantage of the 8th bit for
local processing.  For example, the shell uses the 8th bit for keeping 
track of quoting.

*************************** I NEED YOUR HELP! ***************************

I need to prepare a list of all the places in UNIX (e.g., drivers, library
routines, utilities, etc) that use the 8th bit.  I would also appreciate
any suggestions on how these places can be changed to not use the 8th bit.
If you have any information that might be helpful to me in preparing this
list, I'd appreciate it.

	Thank you,

	Gary Lindgren
	AT&T Bell Laboratories
	190 River Road
	Summit, N.J.  07901

	ihnp4!attunix!gll
	201-522-6170
	x6170  sfsno!gll

------------------------------

Date: 8 Feb 85 06:08:29 GMT
From: "George R. Cross" <cross%lsu.csnet at CSNET-RELAY.ARPA>
Subject: Tektronix 1240 Logic Analyzer

Does anyone have or know of any interesting software to control
the Tektronix 1240 Logic Analyzer through the RS232 port under Unix?

George R. Cross, Computer Science Department
Louisiana State University, Baton Rouge, LA  70803-4020
Phone: 504-388-1495
CSNET: cross at lsu	ARPA: cross%lsu at csnet-relay
BITNET: cscros at nsnccvm	cscros%nsnccvm.BITNET at wiscvm.ARPA

------------------------------

Date: 6 Feb 85 17:29:03 GMT
From: roy at phri.UUCP (Roy Smith)
Subject: F77 -> C routine communication

I have a program which has f77 and C modules (main in fortran,
much of the I/O in C).  On the fortran side I have declared
"character*512 (512)".  If I want to pass this 512x512 array of
characters to a C routine, how should I declare the formal
parameter?
-- 
allegra!vax135!timeinc\
   cmcl2!rocky2!cubsvax>!phri!roy (Roy Smith)
         ihnp4!timeinc/

------------------------------

Date: 11 Feb 85 22:35:11 GMT
From: Michael M. How <OA.HOW at MIT-XX.ARPA>
Subject: LZ compression algorithm

Judging from  recent messages, I see they is some interest in LZ 
compression algorithm. I would like to sum up some of the work I have
done with the algorithm and exchange some ideas about implementation.

Back  in Oct. 1983 , Miller and Wegman from IBM Yorktown gave a talk
about their extensions to the LZ algorithm. They proposed an LRU
replacement to handle dictionary overflow. They also devised a string
extension version of the algorithm. In particular, to form the new
string to insert in the dictionary, concatenate the previous dictionary
match with the match just found. This allows much faster buildup of
dictionary entries. The most interesting part of their work was
applying this to terminal sessions. The algorithm sat between an IBM PC
and a CMS system. Results they claim are 4 to 8:1 compression, so you
1200 baud sessions look 9600. I investigate applying this to a character
oriented system, UNIX in particular. Results of course are good. 

What I am interested is finding other implementation of the LZ compression
and seeing how people handle dictionary overflow, matching. I would be
interested in performance figures. The IBM implementation running on a
3081 compressed 1 million character of English text in less than 20
seconds. My version has compressed about 1000 chars/sec for 90K file on
a VAX/750. In using this for terminal sessions, I would like to improve
the speed on the VAX (the decompression side on the PC just rips).
	Thank you,
	Michael How

------------------------------

End of Unix Technical Discussions
*****************************
-- 
Ronald W. Heiby / ihnp4!{wnuxa!heiby|wnuxb!netnews}
AT&T Information Systems, Inc.
Lisle, IL  (CU-D21)