Info-IBMPC@C.ISI.EDU (Info-IBMPC Digest) (04/28/87)
Info-IBMPC Digest Monday, 27 April 1987 Volume 6 : Issue 30
This Week's Editor: Billy Brackenridge
Today's Topics:
MAJOR Glitch in Turbo Pascal Command.com
Problem with Turbo Pascal and Machine Misbehavior
CALC.E Calculator for Epsilon
xbbs
Honeywell Terminal Emulator
Graphics toolbox for TurboPascal is in SURFMODL
ZStem on Z-248
ERROUT.C
Borland Turbo Basic
SPL
Text-base software
/V in COPY not Same as /V in DISKCOPY
PC/XT Clone BIOS
Driving TURBO PASCAL from a Batch File
Computer Innovations C86PLUS C Compiler V1.01
PS2 Color Resolution Less than Mac II (2 Msgs)
DIAL.SRC from Info-IBMPC lending library
Logo to work with EGA
Getting Information about MS-DOS-Machines Through Networks
HDLC Interface
Running Norton's "SI" on IBM Personal System/2 Model 60
47th Street Mail Order
EGA PrtSc Found PGA PrtSc Wanted and 3279 Emulators
8087-80287 and Equipment Flags
Freemacs List
Secure Xenix for the IBM AT
Wildfire from Software Wizardry and an Alternative
3Com EtherLink Cards PCs and DMA
Tight Loops (2 Msgs)
INFO-IBMPC BBS Phone Numbers: (213)827-2635 (213)827-2515
----------------------------------------------------------------------
Date: Mon, 20 Apr 87 23:40:34 cst
From: zeke!pauls@a.cs.uiuc.edu (Paul Spencer)
Subject: MAJOR Glitch in Turbo Pascal Command.com
I wouldn't particularly blame either DOS or Turbo Pascal. The PC/XT/...
does not provide a way to protect memory from (in this case, system memory)
from misbehaving programs (of any language). I suspect that when your
friend did #2 (#1 isn't particularly bad unless you use do a lot of
allocating in which case you run out of memory...dies with a run time
error) part of the system memory got clobbered (part of DOS remains resident
to do I/O and handle system calls; COMMAND.COM is just the command
interpreter). When you rebooted, everything was cleared out and reset.
Turbo Pascal does not check to make sure that everything you dispose was
actually once "newed". To do so would substantially slow down the execution
of your program. I suppose they could have supplied a compiler switch (like
for range checking) but they didn't.
------------------------------
Date: 23 Apr 87 15:27 PST
From: HLL%SLACVM.BITNET@wiscvm.wisc.edu
Subject: Problem with Turbo Pascal and Machine Misbehavior
This is in response to the query by Bob Brankley of 15 April 1987
about a 'Major Glitch in Turbo Pascal.' He complained of machine
malfunction after exiting Turbo. The problem is almost certainly
what he described as disposing of memory which was not allocated. A
Turbo Dispose plants a Hex 0000 at the address of the variable being
disposed, and it does not care where it might be. Disposing of a nil
pointer will overwrite the interrupts! This has been fixed in the
current version of Turbo, 3.02; it will refuse to dispose of a nil
pointer. A random pointer, however, is an open invitation to disaster.
------------------------------
From: Yuval Rakavy <yuval%humus.huji.ac.il@RELAY.CS.NET>
Date: Tue, 21 Apr 87 12:46:15 jdt
Subject: CALC.E Calculator for Epsilon
Attached below is a pocket calculator for Epsilon.
With this command added to your epsilon, you can very easily define
"smart" keyboard macros such as line-numbering, automatic table generation
etc. without the need to write EEL programs (a thing you will seldom do
for a one time task, and therefore you will perform this task manually...)
You can paste the result of the last calculation into the text. An unlimited
number of variables can be used etc.
I will be happy to get improvements and/or suggestions for such.
The file attached below should be named CALC.E
Yuval Rakavy
Hebrew Univ. of Jerusalem Israel.
[CALC.E has been added to the info-ibmpc EEL lending library. It can be found
on c.isi.edu in the directory <info-ibmpc.eel> and in the eel directory of
our BBS. -wab]
------------------------------
From: Sandy#308@DLX-GATEWAY.ISI.EDU
Date: 04-18-87 07:12:16
Subject: xbbs
Hi there,
I am the author and user of XBBS. You can reach the system by calling
714-828-0288. Unlike other BBS systems, this one runs under Xenix and
can handle as many callers as you have ports/modems. I make available
the full "C source to the code.
I have been trying to make my system known through info-ibmpc without
any luck. I did, however, stumble across your telephone number in
one of the info-ibmpc releases.
My system caters to Unix/Xenix/Msdos users; however, a major part
of the file sections is my bbs software. You, like all others, are
more than welcome to download all of the "C" source. My software
can handle up to 99 bulletin bases, 99 message bases, and up to 99
file bases. Every command, every bulletin base, every message base,
and every file base has a privilege level assigned to it. There are
many additional features in the code such as changing a users privilege
online, online chatting, and viewing activity on a multi-screen or
external terminal.
I can handle the following file protocols: kermit, super kermit (
sliding windows ), xmodem-crc, xmodem-checksum, batch ymodem, ymodem-crc,
ymodem-checksum, ascii, and type. I will be implementing soon: Sea's
protocol, telink, and modem7.
Sanford ( Sandy ) Zelkovitz
XBBS ( Xenix BBS ) 714-828-0288
ihnp4!hermix!ucla-an!alphacm!{root, sandy}
------------------------------
Date: Tue, 21 Apr 87 06:50:25 CST
From: mlw@ncsc.ARPA (Williams)
Subject: Honeywell Terminal Emulator
For our friends at nrl...
The Air Force microcomputer follow-on contract (i.e., the Zenith Z-248
contract) includes a VIP 7700/7800 Emulator -- that's its title. It's
on contract F-19630-86-D-0002 as CLIN 0032AD, part CC-5063-2, and costs
$156. I have no information on its ease of use or performance. It's
got the advantage, though, of being able to use the COM3 port that comes
on the contract Z-248s. I don't know if it'll work on other systems or
not.
Mark L. Williams
(mlw@ncsc.arpa)
------------------------------
Date: Tue, 21 Apr 87 9:13:23 EST
From: Kenneth Van Camp -FSAC- <kvancamp@ARDEC.ARPA>
To: A4422DAB%AWIUNI11.BITNET@WISCVM.WISC.EDU
Subject: Graphics toolbox for TurboPascal is in SURFMODL
>I would need a graphics toolbox for TurboPascal fulfilling the
>following requirements:
>Running with CGA
>Fill with at least 5 different shades of gray (possibly also in colors)
>for closed polygonal regions.
I implemented such an algorithm in Surfmodl, which is released to the public
domain. Full Turbo Pascal source is available from either USC-ISIB or
SIMTEL-20. My Fillsurf routine fills closed polygons of any degree, either
by color number or grey-scale via dithering. The only reason it's not a
totally general-purpose filled surface routine is because I wanted to speed
up the access to Surfmodl's own internal data structures; you'll have to
rewrite it for your own data structures.
--Ken Van Camp <kvancamp@ARDEC.ARPA>
------------------------------
Date: Tue, 21 Apr 87 11:44:20 AST
From: ernst@afit-ab.arpa (Thomas J. Ernst)
Subject: ZStem on Z-248
The ZStem fix for the backspace key has been provided by Robert
Bloom (rbloom@apg-1) is as follows "The problem in ZStem is that
the backspace is a command character to ZStem - even when defining
softkeys. A <bs> deletes the previous character entered including
softkey definitions. To get around this, temporarily redefine
the ZStem <cancel> key, i.e. item #43 on the configure screen, to something
else, then redefine it back as necessary."
Using this method you can redefine <bs> to a control-H, which will do
the desired destructive backspace as desired.
------------------------------
Date: Sun, 19 Apr 87 19:46:22 pst
From: Robert Lenoil <lenoil%apple.csnet@RELAY.CS.NET>
Subject: ERROUT.C
Last week I sent in a (very) small program to redirect stderr to
stdout and then execute its command line, as a response to someone's
posting on patching MASM to write its errors to stdout instead of
stderr. Unfortunately, I goofed: instead of redirecting stderr (2)
to stdout (1), I redirected stdaux (3) to stderr (2). Here's a
corrected version, with some added code for online help. Feel free
to put it in the library.
Robert Lenoil
/* This program redirects stderr to stdout, then executes its command line.
* Written by: Robert Lenoil - 4/87 Compile with: Microsoft C, v4.0
*/
#include <process.h>
main(argc,argv)
char *argv[];
{ if (argc > 1)
{ dup2(1,2); /* redirect stderr (2) to stdout (1) */
++argv;
return execvp(*argv, argv); /* execute command line */
}
/* no command line, print some help */
puts("\"ERROUT command_line\" executes command_line,");
puts(" with stderr redirected to stdout.");
puts("A useful example would be: ERROUT MSC FOO.C;");
puts(" | FIND \") : \" > FOO.ERR");
puts("which would compile FOO.C, writing all compiler error");
puts(" messages (which are all");
puts("of the form \"filename(linenumber) : messagetype number:");
puts(" message\") to the file");
puts("FOO.ERR. Written April, 1987 by Robert Lenoil.");
return 1;
}
------------------------------
Date: 21 Apr 87 16:17:39 PDT (Tuesday)
Subject: Borland Turbo Basic
From: "Larry_Shilkoff.ElSegundo"@Xerox.COM
I recently purchased a copy of Turbo Basic and have found at least one
bug (somewhat critical in my opinion). Borland uses two methods for
formatting single precision real numbers for storage in a random access
file. The functions MKS$ and MKMS$. MKS$ uses an IEEE format while
MKMS$ uses the existing Microsoft format. MKS$ works fine but MKMS$
will store the absolute value only (i.e. using MKMS$(-64) then a PUT
will store 64) making Turbo Basic absolutely incompatible with any
existing MIcrosoft BASIC files incorporating negative single precision
reals.
After discussing this with Borland, they did not seem to be too
concerned about correcting the problem.
caveat emptor
Larry
------------------------------
Date: Tue, 21 Apr 87 20:55:57 EDT
From: Russell Nelson <bh01%CLUTX.BITNET@wiscvm.wisc.edu>
Subject: SPL
SPL is a turkey. When I first heard of it, I was excited. Oooo, a shareware
compiler. Neat! Unfortunately, after I downloaded the thing, I found out
that it was just a preprocessor for BASIC. BASIC needs a preprocessor like
a fish needs wheels. I realize that you have a liberal posting policy, but
isn't there a way to weed out junk like this. Especially since it was a
press release, which are notoriously informationless and hype-filled.
-russ
p.s. feel free to ignore this flame. :-)
[I sort of felt this way as I was editing the press release, but they were
up front that this is what the product did and there was lots of tech
information so I ran it. -wab]
------------------------------
From: sphinx.UChicago!sphinx!goer@oddjob (Richard L. Goerwitz III)
Subject: Text-base software
Date: 22 Apr 87 01:44:10 GMT
Organization: U Chicago Computation Center
In response to some inquiries concerning text-base software, I would like
to offer the following program as a possible solution for those who would
like to have this text-base fully integrated with their word-processor.
The program is called Nota Bene. It's put out by Dragonfly Sofware (212-334-
0445). What it does is allow the user to take papers, notes, etc., and create
for them an index, which may then be searched in any number of ways - title,
keywords, etc. (Boolean logic of course). Academics will find this useful for
bibliographic files, for keeping track of old articles, and what not.
The real advantage to Nota Bene is that, in addition to being a
text-base, it is also the best academic word-processor on the market.
Yes, I have used WordPerfect, Microsoft Word, etc. They all fall
short. It is the most powerful and flexible wp I have ever seen.
Right now I use it for multilingual articles written in English,
Hebrew, Greek, Syriac, etc. The fact that the text-base is fully
integrated into it makes the program doubly useful.
For those who would like to know, Nota Bene is actually a modified version of
XyWrite. Its text-base is, I believe, a daughter of FYI5000.
Needless to say, I have no financial interests at stake here. I'm merely
passing on information that has proven useful to me.
-Richard
goer@sphinx.uchicago.edu
------------------------------
Date: Tue, 21 Apr 1987 23:55 MDT
From: WANCHO@SIMTEL20.ARPA
To: "JOHN R. THOMPSON" <WOOLFORD.THOMPSON@BIONET-20.ARPA>
Subject: /V in COPY not Same as /V in DISKCOPY
John,
Note that the /v option description for COPY is not quite the same as
the /v option for DISKCOPY. The difference is subtle: the /v for COPY
is described as "causes MS-DOS to verify that the sectors written on
the target disk are recorded properly." DISKCOPY's /v "will verify
that the contents of the source and destination disks agree."
The action for the /v option of DISKCOPY is what you would have
expected to see described for COPY. My interpretation is that COPY
simply tries to read what it just wrote, i.e., it verifies that the
write took place, and nothing more - nothing about whether or not the
contents written in the target file actually matched the contents of
what it read from the source file.
--Frank
------------------------------
Date: 22 Apr 87 08:44:00 EDT
From: "V703::S_DANIELS" <s_daniels%v703.decnet@nusc>
Subject: PC/XT Clone BIOS
Last week someone asked for info on the various BIOS' for clones.
I am preparing somewhat of a brief overview of the few I have had
experience with (ERSO, Golden XT, etc). Have patience, it will soon
be done - I don't want you top think your request is being ignored.
SCOTT
------------------------------
Date: 22 Apr 87 08:43:40 PDT (Wednesday)
From: Halsema.es@Xerox.COM
Subject: Driving TURBO PASCAL from a Batch File
In reply to Keith Redwill:
TURBO can be driven from a batch file. A PC Magazine Utility called
KEY-FAKE, and a patch to TURBO (published in PC Magazine, January 13,
1987, page 314) will allow you to do what you want.
The patch is needed because TURBO will stop compiling if it detects a
character in the keyboard buffer.
Enjoy,
--Ian
------------------------------
Date: Wed, 22 Apr 87 13:41 PST
From: <IVAN%TRIUMFCL.BITNET@wiscvm.wisc.edu>
Subject: Computer Innovations C86PLUS C Compiler V1.01
Some time ago a query was posted regarding Computer
Innovations C86PLUS C Compiler. Below is a (relatively) brief
summary of my experiences so far. (Note: TEKEM is a Tektronix
40xx/VT640/CIT467 grafix terminal emulator).
When I first received my C86PLUS compiler, I was using a PC
Clone with 640K, V20/8087 chips and a 20 Mb disk. The package
consists of two manuals (user guide and library reference) and 6
diskettes. I immediately ran into problems trying to compile the
"sieve" test program, with all sorts of weird errors as I tried
various combinations of FILES and BUFFERS in the CONFIG.SYS file.
These errors included the whole gamut of compiler errors, (WRITE,
CORE, etc.) and a continuous Print-Screen trigger! I eventually
removed my two TSRs, DOSEDIT and SEARCH, and things proceeded a
little better. (Soon afterwards I received MS-FORTRAN V4.00 and
found the same problems. It appears that SEARCH will not co-exist
with overlaid executable -- I have changed to DPATH and the problem
seems to have disappeared.)
However, I was still not able to link the sieve program as
there was an undefined external at link time! I found that I could
link with the DOS linker and run the program with the following
results: C86 V2.20M -- compile & link= 37", run=14.35"; C86PLUS V1.01
(large model) compile & link=2'38", run=5.10". I finally determined
that the problem with the undefined external was that there was a
module missing from the small (and compact) libraries! Attempting to
compile & link hello.c under the 4 models resulted in: small -
undeclared external; compact - undeclared external; medium - fixup
error in linker; large - successful. Thus the medium library
contains the missing module, but has at least one module compiled
under the wrong model.
Apart from those problems, I also found that the
documentation is poor and extremely sparsely indexed. The books are
dated October '86 while the compiler is March '87, and there appears
to be little correspondence between the two. This is especially true
of the documentation of the compiler flags. The "read.me"
documentation suggests compiling the sieve test with -Fm and -F1000
flags, which of course are not documented, while the documentation
for the -E flag says that it will run the pre-processor only; instead
the pre-processor output is sent to stdout, and compiling and
linking continues.
At this point I sent a letter and SPR off to Barbara
Eberhardt and am still awaiting a response. In the meantime, my XT
was repossessed and replaced by an AT clone, Phoenix BIOS, 8MHz,
1-wait- state, 80287, 20Mb disk, 512K system RAM and 512K extended
RAM. I set it up with the extended RAM used as a RAM-disk. This
machine runs the sieve in about 1.4". Having nothing better to do on
Sunday, I loaded in C86PLUS and tried to compile TEKEM. This
consists of 6 C modules and 3 assembler files. Under C86 V2.20M,
using the RAM-disk for intermediate steps, the whole MAKE cycle takes
about 3 minutes on the AT.
Again, I had problems sufficient to last from midday to
midnight. Firstly, it was SLOW. Using the D: disk for temporary
storage, just ONE of the C modules (472 lines, 11.5K) took 10.5 mins
to compile! This stretched out to 20 mins if I used the hard disk
for temporary files, with the access light continually lit. I could
only compile 3 of the modules on first pass, and had to split the
other three up (lack of memory was a problem -- MAKEing didn't work
due to the extra overhead).
I had continual problems with the compiler hanging, or error
messages such as "PANIC: compilation aborted", and also found that
even with successful compilations I got undocumented errors like
"../subs100.c(341) Unused temporary, Scheme 10." In addition, more
often than not successful compilations left lost chains of 3-400K on
the D: disk (or hard disk if I was using that). Finally I was able
to track down the cause of the compiler's hanging, by using the hard
disk for temporary storage and using lack of access to indicate a
hang, and getting a few hints from error messages that appeared when
I used the -E switch (but not without!). I came across two code
segments that reproducibly hung the compiler, even when distilled
into the two "programs" below. C86 V2.20M and VAX-11 C both accept
these fragments without complaint. Sometimes these sources hang the
compiler so well that Ctrl-Alt-Del won't work.
/* test.c */
#include <stdio.h>
FILE *p;
int flag;
main()
{
if (p != 0)
fclose(p);
p = flag =0;
exit();
}
/* test2.c */
#include <stdio.h>
int p();
int dx_reg;
main()
{
dx_reg=p;
exit();
}
So, there seems to be a problem with assignments from one
storage class to another. If it was going to be that picky, I would
prefer that it complained rather than hanging!
Anyway, I programmed around these two problems (using a
union in the second case -- this to get the address of an interrupt
routine), and then spent a few hours converting the assembler files
to conform to the new conventions (which are identical to MS-C). It
only took 2 passes to link (I forgot to include an extern for a mod I
made) and got the same undeclared external. Then the bloody
program hung the computer when I tried to run it!
So the upshot is: I do not recommend C86PLUS V1.01. I hope
that Computer Innovations will respond to my complaints and fix all
the bugs. I hope they can speed it up a bit too -- I estimate 80
minutes total to MAKE TEKEM on the AT (cf 3 for V2.20M) and that
translates to about 6 hrs on a PC (assuming a large RAM-disk)!
Date: Sat, 25 Apr 87 17:36 PST
I reported earlier on problems I was having with Computer Innovations
C86PLUS C compiler V1.01. I have since found out why my program bombs
when I try to run it -- the compiler generates erroneous code which causes
a test to fail and leads to an illegal call to free() as well as a corrupted
stack. Output from the compiler with -Fa (generate assembler file) is:
;55:{ if (*(ext_code[0][i]) != '\0')
mov si,word ptr _i@1[bp] ; 000f 8b 76 fe
sal si,1 ; 0012 d1 e6
mov di,offset dgroup:_ext_code ; 0014 bf 0000R
add di,si ; 0017 03 fe
mov di,word ptr [di] ; 0019 8b 3d
mov al,byte ptr [di] ; 001b 8a 05
xor ah,ah ; 001d 33 e4
or ax,ax ; 001f 0b c0
mov word ptr T2@2[bp],si ; 0021 89 76 a6
jz L18 ; 0024 74 13
However, when I check this portion of code with SYMDEB, I find:
1670:1B9E 8B76FE MOV SI,[BP-02]
1670:1BA1 D1E6 SHL SI,1
1670:1BA3 BF7820 MOV DI,2078
1670:1BA6 03FE ADD DI,SI
1670:1BA8 8B3D MOV DI,[DI]
1670:1BAA 8A05 MOV AL,[DI]
1670:1BAC 33E4 XOR SP,SP
1670:1BAE 0B.C.0 OR AX,AX
1670:1BB0 8976A6 MOV [BP-5A],SI
1670:1BB3 7413 JZ _init_codes+39 (1B.C.8)
1670:1BB5 8B7EFE MOV DI,[BP-02]
Thus the code generated for XOR AH,AH is wrong and, as you can see,
this affects both the outcome of the test (ah is non-zero) and the value
of the stack pointer.
------------------------------
Date: 22 Apr 87 12:08:08 PDT (Wednesday)
Subject: PS2 Color Resolution Less than Mac II
From: "Thomas_R._Moenter.rochX2"@Xerox.COM
According to actual IBM-supplied announcement literature, the PS-2
graphics are not as spectacular as your note indicates. In 640x480 mode
only 16 colors are available. The 256 color mode is in a PC-compatible
resolution of 320x200. This is markedly inferior to the Mac II ability
of 256 colors at full resolution. (640x480 in standard Apple dress)
------------------------------
From: "Thomas_R._Moenter.rochX2"@Xerox.COM
Date: 27 Apr 87 05:50:06 PDT (Monday)
Subject: PS2 Color Resolution Less than Mac II
I'm fairly certain that the 640x480 mode on the PS-2 is non-interlaced.
I didn't notice any flicker on the displays I saw. The demos carefully
select objects with a few shades of a single color, i.e. a cut tomato
and a green pepper, so that the fact they only support sixteen colors
can be hidden. Sixteen colors is fine for pie and bar charts but
woefully inadequate for rendering color images. The Mac II with full
support in ROM for up to 32 bit color will spawn incredible image
processing and display software/hardware within the coming year.
Supermac Technologies has announced 19" screen and 8 bit (256 color)
controller card that will lead the wave of future imaging stuff for the
Mac.
I suspect that a VGA board may be made available at some point for the
base PC, however much depends on actual demand for PS-2's. IBM wants you
to buy PS-2's not upgrade your old junk.
------------------------------
Date: Wed, 22 Apr 87 19:33:25 est
From: mm59#@andrew.cmu.edu (Mark L. Milliman)
Subject: DIAL.SRC from Info-IBMPC lending library
Today I ftp'ed DIAL.SRC from the lending library, then tried to run the shell
to create the MSC source files. In ftp I used a standard ASCII transfer.
Next, I removed the header from the script up to the first comment. Then I
ran it using /usr/sh. It spit out the following errors:
shar: error transmitting 'dial.doc' (should have been 4481
characters)
shar: extracting 'dial.c' (10750 characters)
shar: error transmitting 'dial.c' (should have been 10750
characters)
shar: extracting 'ibm_comm.asm' (6773 characters)
shar: error transmitting 'ibm_comm.asm' (should have been
6773 characters)
shar: extracting 's_msc.mac' (1237 characters)
The actual sizes of the newly created files are:
10866 Apr 22 20:19 dial.c
4443 Apr 22 20:19 dial.doc
6901 Apr 22 20:19 ibm_comm.asm
1237 Apr 22 20:19 s_msc.mac
I really do not want to get into and hack the shell script unless I it is
necessary. Was the program improperly put into the library or is there
something wrong with it?
Thanks for your help,
--Mark Milliman
UUCP: ..!ihnp4!hocad!mlm
..!ihnp4!cmucspt!ampere!mlm
ARPA: mlm@ampere.ece.cmu.edu (most reliable)
[I am afraid this got messed up removing the uuencoded executable files.
I did it by hand as I don't speak Unix. This might have messed things
up a bit. -wab]
------------------------------
Date: Thu, 23 Apr 87 01:10:10 EDT
From: Michael Friendly <FRIENDLY%YORKVM1.BITNET@wiscvm.wisc.edu>
Subject: Logo to work with EGA
IBM Logo (developed by Logo Computer Systems, Inc) is the only full Logo
I know of for the PC which works with the EGA. Unfortunately, LCSI
sold the marketing rights to Big Blue, so the price ($250.US) is
somewhat out of line with recent products. IBM doesn't even offer
educational discounts or site licenses for educational settings.
On the other hand, IBM Logo is far superior to DR Logo in most
respects, in spite of some very nice advanced debugging features
in DR Logo. The latter locks you in to its own operating system
and implements recursion inefficiently so that you are forced to
avoid recursion where speed matters.
You might look into LogoWriter, a new LCSI product designed specifically
for lower grades, with generous site licensing and a great deal of
teacher support material. LogoWriter adds some word processing
capabilities to Logo, but at the expense ( in the current release )
of some of the more advanced features of Logo. I do not know whether
EGA support is provided.
------------------------------
Date: Thu, 23 Apr 87 12:54:01 MEZ
From: IWI7B3%DERRZE1.BITNET@wiscvm.wisc.edu
Subject: Getting Information about MS-DOS-Machines Through Networks
A lot of people requested the PC FILELIST I announced a few issues
ago. But I am sorry to tell you that my local mailer at
DERRZE1.BITNET and even some gateways sometimes had problems
delivering all the files sent out by me to the people who requested
the file even using the newest versions of CROSSNET. So I have to
inform those few people who are still waiting for the files that I
have tried to transfer the files to you but didn't succeed. Please
request them once more and let's hope that it will work then.
On the other hand: Which PC library or file server would like to
install the files on its public disks for easy download by any
interested people ? For information about the necessary disk space
feel free to contact me directly and then I will send you the
FILELIST. Hope this helps for this time.
--Thorsten Glattki
Dept.of Computer Science, University of Erlangen-Nuernberg, West-Germany
Netaccounts: IWI7B3@DERRZE1.BITNET, glattki@faui70.UUCP, tnglattk@fauern.UUCP
(+49)9131/857020 (Uni) & (+49)9131/57747 (home)
Dept. of Computer Science,University of Erlangen
Institute for Communication systems,PAP-Project
Martensstrasse 3
8520 Erlangen
------------------------------
Date: Thu, 23 Apr 87 07:56:49 EST
From: mcgurrin@mitre.ARPA
Subject: HDLC Interface
I am looking for information on any boards which provide an HDLC interface
for a PC/XT and/or AT. I've seen many SDLC products listed, but no word
on HDLC. Any help would be appreciated. Please send replies directly
to me since I am only an occasional reader of this group. Thanks!
------------------------------
Date: 23 April 87 16:43-PDT
From: KLASSEN%UVVM.BITNET@wiscvm.wisc.edu
Subject: Running Norton's "SI" on IBM Personal System/2 Model 60
SI reports that the machine has 640K (this one has 1024K),
and a factor of 10.3 times faster than the standard.
------------------------------
From: <iuvax!ndmath!ndcheg!evan@seismo.CSS.GOV>
Date: Mon, 20 Apr 87 12:10:10 EST
Subject: 47th Street Mail Order
I too have been inconvenienced by 47th Street Computer.
However, I've never dealt with them through the mail. Two years ago,
I purchased a 10MB tape drive for the IBM-PC from their Wall Street
store. The thing short-circuited the whole PC. 47th declined to
take it back without testing it in my PC first. Translated... they
wanted me to bring the whole damned computer into their store. It
was only after several heated discussions with their manager that
they agreed to take the tape drive and test it in one of their PC's.
Of course, they then found out that the drive was broken and replaced
it. Since then, I have only purchased from them as a last resort,
and never purchased hardware; only software from large software
companies.
Evan Bauman
University of Notre Dame
..seismo!iuvax!ndmath!ndcheg!evan
gkn3m2@irishmvs (bitnet)
------------------------------
Date: Thu, 23 Apr 87 09:38 ITA
From: Marco Lumachi <POLICDC%IMICLVM.BITNET@wiscvm.wisc.edu>
Subject: EGA PrtSc Found PGA PrtSc Wanted and 3279 Emulators
If you can reach a 3852-002 IBM color ink-jet printer, you can reach its
driver disk. We have two version (1.3 & 2.0) of these drivers. By them you can
print your screen on a 3852 if you have a CGA, a PCjr or an EGA. With release
2.0 drivers you can rotate and magnify the picture (only for CGA) and change
the colors from screen to printer (all PCs and GAs). The driver will not work
on the old graphix printer (i.e. 5151) but, maybe, on the Pro-Printer. This
due to some ESC codes, but I do not have a Pro-Printer so I cannot do the
test.
Now I put a question:
PrtSc for PGA (!).
Has anyone something about ?
Marco Lumachi, MLUMA at IMIPOLI (EARNET)
3279 Emulator:
In our EDP department we use two kind of adapters and emulators. The
first is the 3278/79 adapter (HW) and the 3278/79 Emulation Control
Program (SW). It's a long adapter. I think it is old and here, in
Italy, IBM does not sell it more. With the program you can establish
a session with the host and up/download text files (i.e. ASCII or
EBCDIC files). You can also use one program in the DOS screen (flip
with Alt-ESC) if the program does not make strange I/O operation,
disable interrupts and so on (the pc-host link is synchronous).
The other is the 3270 half size adapter and the PC3270 Emulation
Program. We have rel. 2.0, but rel 3.0 is available. You can do all
the tasks of 3278/79 adapter and more. If you have a PC Network you
can configure a PC as gateway between the net work and the host.
Then, from other PC in network (max. five sessions at the same time)
you can use the host. With our old-timer PCs and PC Network (not
token ring) the sessions are slower than normal 3178 terminal, but
the use of the host from the PC is worth the decrease of speed: no
duplicate terminal, no more cables, easy file transfer .... The PC
3270 Emulation Program supports also a SDLC card and a modem, but we
do not try to use this feature.
Marco Lumachi MLUMA at IMIPOLI (EARNET)
------------------------------
Date: Fri, 24 Apr 1987 14:59:02 ULG
From: Andre PIRARD <A-PIRARD%BLIULG12.BITNET@wiscvm.wisc.edu>
Subject: 8087-80287 and Equipment Flags
To: Erich Neuwirth <A4422DAB@AWIUNI11.BITNET>
Machine equipment is determined in a model dependent way by the BIOS
ROM during power-on. The only valid method for a program to obtain
this information is via interrupt 11 returning equipment flags to
register AX. Coprocessor presence is indicated by bit 1. The XT
BIOS looks at switches on the motherboard, the AT BIOS looks at the
equivalent flags in the TOD clock CMOS RAM. These can be modified by
a program called SETUP on the diagnostics diskette delivered with
the AT. Coprocessor presence is determined automatically by SETUP and
not a question it asks. Both machines *normally* store the equipment
flags at 40H:10H where they are fetched by interrupt 11. Some
(faulty except for diagnostics) programs may look at the hardware
directly and report reality. Others play the game and report what
they're told. Hope this can help. As a last resort, you can look at
what your interrupt 11 does and possibly patch location 40:10, but
you normally should use SETUP or your clone equivalent.
------------------------------
Date: Fri, 24 Apr 87 10:17:19 EDT
From: Russell Nelson <bh01%CLUTX.BITNET@wiscvm.wisc.edu>
Subject: Freemacs List
There is now a mailing list for Freemacs, the PD (actually copylefted)
programmable editor for the IBM-PC and Z-100. To sign up, send a mail
message to LISTSERV at CLVM. If you are on bitnet, you can send a
direct message. The list server command, whether sent by mail or
message, should be "SUBSCRIBE FREEMACS human name". You will receive
notification that you are signed onto the list.
For example, I would subscribe using:
tell LISTSERV at CLVM SUBSCRIBE FREEMACS Russell Nelson
whereas a Internet user would send mail:
SUBSCRIBE FREEMACS Internet Q. User
The help command will return a help screen to you via messages or mail.
-russ
GEnie: BH01
BITNET:BH01@CLUTX
uucp: decvax!sii!trixie!gould!clutx!bh01
------------------------------
Date: Sat, 25-Apr-87 14:31:53 edt
From: David Farber <farber@UDEL.EDU>
Subject: Secure Xenix for the IBM AT
Program Offering - IBM Federal Systems Division, 4/14/1987
IBM Personal Computer
B2 Secure XENIX
Version 1.0
The IBM Personal Computer B2 Secure XENIX Version 1.0 Program Offering
is a multi-level secure operating system for the IBM Personal Computer
AT which utilizes the advanced features of the 80286 processor. It is
a multi-user, multi-tasking secure system, designed to meet Department
of Defense Trusted Computer System Evaluation Criteria for a B2 level
security rating. The product is currently under evaluation by the
National Computer Security Center. In accordance with the applicable
government regulations, the customer is responsible for ensuring
compliance with the appropriate security standards for its particular
operating environment. IBM PC B2 Secure XENIX Version 1.0 contains
many functional and security enhancements while maintaining binary
compatibility with programs developed under IBM Personal Computer XENIX
Versions 1.0 and 2.0.
The IBM PC B2 Secure XENIX system is comprised of three components. A
Secure Operating System is the base component and a prerequisite for
the other components. Included in the Secure Operating System is the
Secure Kernel which controls and allocates all the resources of the
computer according to least privilege security principle, enforces
mandatory and discretionary security policies, performs strict user
identification and authentication, and generates audit trail and
accounting records. Over 150 system commands and three shell command
interfaces are provided with the Secure Operating System.
The second component is the Software Development System which contains
a large set of application software development tools, including the C
programming language. Together, the C language and compiler, 80286
Macro Assembler, interactive debugging, productivity tools and other
programming utilities provide a complete environment for the programmer.
The third component is the Text Formatting System which provides
high-level formatting macros for document preparation in conjunction
with nroff (a formatter for printers) and troff (a formatter for
photo typesetters). Additional formatting facilities are also provided.
Planned Availability: 1 July 1987
One-time charge: $1,868.00
IBM Personal Computer
Compartmented Mode Workstation System
Version 1.0
The IBM PC Compartmented Mode Workstation System capabilities include:
- The ability of the operating system to protect itself and data from
modification.
- The ability to run application programs in single and multiple
compartments.
- The ability to run multiple processes at the same or different
compartments.
- The ability to maintain data and program labeling according to DIA
security requirements.
- Windowing, whereby all users access the system via multiple,
independent windows displayed on the screen, each potentially
associated with a different set of processes in the same or different
compartments.
- The ability to transfer data between windows.
Planned availability: 15 July 1987
One-time charge: $2,568.00 (includes B2 Secure XENIX)
------------------------------
Date: Sat, 25 Apr 87 17:52:14 pst
From: tweten@ames-prandtl.ARPA (Dave Tweten)
Subject: Wildfire from Software Wizardry and an Alternative
Longer ago than I prefer to admit, I inquired to this digest about
Wildfire, from Software Wizardry (OK, so it was the June 7, 1986 issue).
Wildfire had just begun to be advertised as a speedup kit for
Heath/Zenith 150/160 series 4.77 mHz PC compatible computers. I
asked for any experience reports, and offered to relate what I learned.
Nobody had any experience to relate. When I broke down and ordered the
$250 modification I found out why. Software Wizardry wasn't shipping.
Software Wizardry advertised higher speed than Heath/Zenith's 8 mHz
machine. No Wildfire article or advertisement I have seen tells what
clock speed it uses, but my examination of the board's picture on the
cover of the March-April issue of Sextant suggests to me that it uses
the NEC clock chip with a 14.3181 mHz crystal for high speed and a 10
mHz crystal for low speed. Since the NEC chip divides by 2, instead of
Intel's 3, that would give 7.16 mHz high speed, and 5 mHz low speed
operation. Presumably, they depend upon the NEC V20 chip's higher
performance to cover the rest of the distance to the Heath/Zenith 158's
8 mHz 8088 performance. I should emphasize that this is an unconfirmed
guess.
The June, 1986 issue of REMark (the Heath/Zenith Users' Group magazine)
had a construction article featuring a kit by Dante Bencivengo. It
seemed to be similar to Wildfire, and claimed to be able to speed up the
processor to 6.67 mHz (a 40% speed-up), or perhaps 7.38 mHz (a 55%
speed-up). That seemed reasonable, in light of the fact that it cost
only about $40 as compared with Wildfire's $250. Besides, I already had
an NEC V20 chip, so I didn't need another. I ordered it, figuring that
I'd cancel my Wildfire order if it worked.
It did, sort of. With no problem at all, its daughter board and chip
replacements accelerated my Z-151 to 6.67 mHz. As described in the
REMark article, the kit had two problems. When UPS delivered it, a
third problem became apparent.
The article indicated that the kit disabled the Ctrl-Alt-Del warm boot,
because it used an R/C timer to control the relationship between the CPU
clock and the time-of-day clock (Wildfire seems to have this problem
too). In high-speed mode, adjusting the relationship is required for
the boot diagnostic's clock test to work. As the article pointed out,
the kit contains a hardware reset button (so does Wildfire), so
inability to use Ctrl-Alt-Del is not disastrous. By the time the kit
arrived, this problem had been solved by altering the board wiring so
the CPU board's "INT" diagnostic LED controlled the time-of-day clock
speed. The R/C circuit was gone.
The kit also required the user to switch to 4.77 mHz in order to do
floppy disk formatting or verification. It seemed to me that this ought
not to be necessary, since the "INT" LED is software controlled. If
turning on the LED can adjust the clock at boot time, why not also use
it during floppy disk operations which require 4.77 mHz operation? The
problem with this approach was that the kit SPEEDED UP the TOD clock
instead of SLOWING the CPU clock. I called Bencivengo to ask about this
design choice and found out that a second version, which worked the way
I wanted, was on the way. I got him to send me a bare new circuit board
(for $8), and canceled my Wildfire order. Incidentally, Bencivengo is
only promoting the fully assembled version of this improved design, at
$125.
The third problem is that generally available copies of the 8288 bus
controller chip used on Heath/Zenith's CPU board are not up to 7.38 mHz
operation, contrary to what the REMark article claimed. They produce
slightly out-of-spec bus timing. Of the Heath/Zenith provided boards in
my machine, only the memory board cares about this. Bencivengo has
attacked the problem by creating another daughter board which replaces
the 8288 and costs $32, assembled. This board is not required if you
are satisfied with 6.67 mHz operation; the AMD D8288 can handle 6.67 mHz
with a heat sink. Unfortunately, my Tall Tree Systems Jram-2 board
can't handle 7.38 mHz, so I'm sending the daughter board back to
Bencivengo for a refund (he's very good about customer relations).
Bencivengo's documentation says that the two new boards, together,
permit 8 mHz operation. Anyone who wants to try it should be aware that
the memory board's delay line DIP has to be removed. That's no problem
with Heath boards because it's socketed. Unfortunately the delay line
is soldered into the Zenith boards. If you use the right tools, you CAN
unsolder the Zenith delay line and socket it, but this is not a
modification for the faint of heart. Also, Bencivengo's information
indicates that 8 mHz operation requires a terminate-and-stay-resident
program which slows the CPU whenever floppy disk operations are
performed.
There have been several magazine references to both upgrades, in the
Heath/Zenith related press. Some Wildfire references are listed below,
in chronological order:
"Mainstream Computing" column, Joseph Katz, REMark, volume 7,
issue 11, November 1986, pp. 15-19
"Wildfire: A Review", Kevin Lerch, REMark, volume 7, issue 12,
December 1986, pp. 65-66
"Wildfire Speeds Up the '150 or '160 to Match the '158", Charles
Floto, Sextant, issue 27, March-April 1987, pp. 8-11
References to the Bencivengo kit are all from REMark:
"H-150 Speed-up Modification", D. Bencivengo, volume 7, issue 6,
June 1986, pp. 55-59
"Letter To D. Bencivengo On His June '86 Article 'H-150 Speed-up
Modification'" letter to the editor, Donald G. Harley,
volume 7, issue 12, December 1986, p. 7
"H-150 Speed-up Mod" letter to the editor, D. Bencivengo,
volume 8, issue 3, March 1987, p. 81
"REMark, June 1986, D. Bencivengo" letter to the editor, R. J.
Maskasky, volume 8, issue 4, April 1987, pp. 9-10, 83
"H-150 Speed-up Modification Update", Dante Bencivengo, volume 8,
issue 4, April 1987, pp. 51-53
The final step I took was to write my own terminate-and-stay-resident
program to automatically switch to slow speed when a floppy disk verify
or format call is made to the BIOS. It results in a system every bit as
convenient at high speed as my machine used to be, at low speed. I'm
not sending it to the Info-IBMPC source library because it seems to be a
special interest item. If anyone's interested in it, send me mail.
So there it is. My experience with Wildfire is that I canceled my
order after about three months, and got an alternative. The alternative
is cheaper and slightly more convenient (Ctrl-Alt-Del works), although
the possibility exists that Wildfire might have been able to drive my
Jram-2 board at a 7.4% higher V20 clock speed. I have no relationship
to either vendor, except as a (would be) customer of varying degrees of
satisfaction. If you wish to contact either, their addresses and phones
are:
Software Wizardry Dante Bencivengo
1106 First Capitol Drive P.O. Box 234
St. Charles, MO 63301 Wyandotte, MI 48192
(314) 724-1738 (313) 484-4866 (eves. 7-10 PM)
------------------------------
Date: Thu, 23 Apr 87 17:14 EST
From: BEAME%MCMASTER.BITNET@wiscvm.wisc.edu
Subject: 3Com EtherLink Cards PCs and DMA
Hello again,
Since I last sent the message about DMA being slow on my
TI/BP I have tested the DMA vs Loop on a real IBM PC.
I had the PC send 1000 packets of the selected size and
had my Excelan LANalyzer monitor the time between the first and
last packets transmitted . I was able to achieve accuracy of 1/2 ms.
Here are the Results:
Time given is the number of seconds taken to transmit 1000 packets.
IBM PC: (real one)
60 100 138 600 1500 Bytes
----------------------------------------------------------------
Loop 0.891 1.293 1.657 6.207 15.085
DMA 0.634 0.767 0.885 2.355 5.243
----------------------------------------------------------------
This shows that on an IBM PC DMA is about 3 times faster then a tight
loop.
Zenith PC: (8Mhz)
1500 Bytes
--------------
Loop 10.461
DMA 4.561
--------------
TI/BP: (Turbo on {No wait states on memory})
1500 Bytes
--------------
Loop 6.509
DMA 10.745 !!!!
DMA 10.431 (Turbo OFF)
When I first tried this experiment I had the PC send the packet to
it's own ethernet address. Guess what? With packet sizes above 187 bytes
every packet had a CRC error and was received as being about 187 bytes!
This also occurred when transmitting to the broadcast address.
I think the 3Com card is attempting to receive it's own packets and
screws up the CRC calculation in the process. Maybe telling the board
not to receive any packets might fix the problem, but you would have
to reset the board after transmitting and would thus lose more packets
then it already does.
I hope no protocol requires a broadcast packet greater then 187 bytes.
- Carl Beame
BEAME@MCMASTER
[These messages are from the pcip mailing list, but are relevant -wab]
------------------------------
Date: Thu, 23 Apr 87 17:56:17 AST
From: "John A. Shriver" <jas@monk.proteon.com>
Subject: Tight Loops
A tight loop is a known loser on a 8088, as it guarantees that every
other instruction will not be in the prefetch pipeline. You have to
unroll loops (say 8 moves per compare & branch) to get any sort of
speed out of them.
Of course, loops beat the socks off 8 bit DMA on a AT.
------------------------------
Date: Thu, 23 Apr 87 22:16:04 EDT
From: Russell Nelson <bh01%CLUTX.BITNET@wiscvm.wisc.edu>
Subject: Tight Loops
True, a tight loop is a loser. Unrolling this way may be obvious, but then,
it may not...
mov cx,count ;may be prime i.e. not a power of 2
a: in al,dx
stosb
dec cx
jz b
in al,dx
stosb
loop a
b: ret
Another possibility is to insist upon having a V20. They're cheap, and
give you another 10% plus the string i/o instructions. In other words,
assume that they have insb and outsb and write your code accordingly.
After all, they have to open up the machine to install the Ethernet
card anyway... :-)
-russ
------------------------------
End of Info-IBMPC Digest
************************
-------