Info-IBMPC@C.ISI.EDU (Info-IBMPC Digest) (05/20/87)
Info-IBMPC Digest Tuesday, 19 May 1987 Volume 6 : Issue 37
This Week's Editor: Billy Brackenridge
Today's Topics:
C++ Availability
Copyright status of ARC, LZW, and COMPRESS programs questioned
A Fix Please for Backspace and Long Command Lines
IBM's PC TCP/IP
How to Detect an EGA?
720K 3.5" floppy with DOS 3.3
Redirection in Batch Files (2 Msgs)
Z-248 (8 Mhz 0 wait state) memory upgrade
Problems Repacking DOS Partitioned Split Disks with REFORMAT
DIAL.SRC unpacks OK
Contribution of GRAPH
Hercules & CGA Incompatibility
Software for the Handicapped
Mailing List for Handicapped
PROTECT.ASM problem
Turbo C
S & S Wholesalers BEWARE!
Bug in Microsoft FORTRAN V4.0 Compiler
fix to DU
Fastcard
Defeating Model 339 ROM POST
Opening Floppy Drive Door Crashes DOS 3.2
Today's Queries:
Query about AT&T PC 6310
IBM Voice Communications Option Sought
WANG Plotter
Sort Programs
Text Files: DEC <-> PC
5 Mhz Math Coprocessor on Turbo xt
Telecommunication Outside the U.S.
Microsoft Cobol is 1974 Standard
Dbase, Lotus information
Pop-Up Dictionary
286 Accelerator boards for XT
Int 32 and 52
Pinouts for the M24 computer.
shipping a pc overseas
EMM and DMA
More on the ANSI question
INFO-IBMPC BBS Phone Numbers: (213)827-2635 (213)827-2515
----------------------------------------------------------------------
Date: Wed, 13 May 87 21:03:13 PDT
From: larry@Jpl-VLSI.ARPA
Subject: C++ Availability
The current issue of Computer Language discusses the current
implementations of C++ for MS-DOS machines.
Larry @ jpl-vlsi.arpa
------------------------------
Date: Monday, 11 May 1987 21:56-MDT
From: "James A. Woods" <jaw%aurora.uucp@BRL.ARPA>
To: info-micro@BRL-VGR.ARPA
Subject: Copyright status of ARC, LZW, and COMPRESS programs questioned
# "Don't compress that dwarf -- hand me the pliers!" -- after Firesign Theatre
> 2) The original Ziv-Lempel method is patented (#4,464,650 -- Willard
> Eastman, Abraham Lempel, Jacob Ziv, Martin Cohen) assigned to Sperry
> Univac (now Unisys). Since the Welch modifications are to this
> method, I would think that some sort of license agreement from Unisys
> would be necessary (this is really only a practical problem for
> commercial customers). Does such an agreement exist?
>
Professor Lempel once telephoned me to praise the existence of a
public domain implementation of the LZ algorithm. Though I can take credit
only for the encoder hashing method currently used in 'compress', as well
as its "block-adaptive" table reset strategy (we remain indebted to Spencer
Thomas of the Univ. of Utah who gave USENET the basic framework), I'll
repeat here a comment relayed to me after a Lempel lecture at HP Labs.
The story goes like this: apparently the Welch paper came to the
light of day only after Sperry Research Labs was disbanded, this occurring
long before the Burroughs acquisition. Supposedly, discoveries revert
to the general public when a lab ceases to exist. Note this is *not*
the same as the situation engendered by the recent asset transfer from GE
to SRI of the RCA David Sarnoff Labs.
In any event, the Welch implementation (Computer, vol. 17, #6,
1984) only claimed that the presented "hardware" string table
creation and access method was "Sperry proprietary". Details of any
software-based code/index storage scheme in existence at Sperry were
deliberately left fuzzy in the paper.
Since patents cover only "apparatus", no one is making claims for
any of the algorithmic variants of LZ, of which there are many (see below).
As for the copyright status of 'compress', Thomas and I (who both work at
public institutions) have signed (meaningless?) waivers not only to UCB
for the 4.3 distribution, but to Hewlett Packard for inclusion in their
own Unix release. The latter is most ironic, since HP retained Lempel
as a consultant for a year on sabbatical leave from Technion in Israel,
where he was chairman of the computer science department.
ARC is another matter, of which I know little. It is fine by me
if someone sells a value-added 'compress' (you'd pay for the packaging
and "support"). Other companies sell the Unix LZ as part of a product
(the Talaris 'troff' software includes compressed fonts this way). Now
I hear that Dan Robinson of Telebit (our friendly neighborhood 18 kbps
modem supplier) has valiantly jammed compression into the modem ROM,
adding a few tricks of his own, no doubt. Speaking again only for myself,
it doesn't matter even if raw unadorned 'compress' were sold for a megabuck --
word would get around very quickly that it's available free from other
sources.
LZ algorithms are not the be-all-end-all of data compression techniques.
They don't particularly work well (unmodified), for digital sound processing
or color picture reduction, for example. Many variants employ equally many
time-space tradeoffs, with software implementations using data structures
ranging from binary trees, to "trie forests", to hash coding, to a direct
sparse array access exercise (for multi-megabyte machines) I posted to
USENET back in 1984/5. Software work continuing at the Univ. of Calgary
should be mentioned, where Tim Bell claims a 5-10% rate improvement (for
ASCII-only input, alas), and unfortunately using an encoder which runs
a hefty order-of-magnitude slower, limiting application. (See his IEEE Trans.
Comm. paper of Dec. 1986, which oddly sidesteps direct comparisons with
'compress'). Also, many ad hoc and not-so-ad hoc methods have been offered
to squeeze data, including the very involved Markov schemes of Cleary
and Witten, and the nouveau self-adaptive splay-tree amortization
algorithms of Bentley and Tarjan.
I could go on, but close by indicating that though optimal data
compression in general is unsolvable in the Turing sense, and though
many problem subclasses are NP-complete, the beautifully simple, linear,
and general method of Ziv and Lempel is hard to improve upon, and certainly
affords many approaches not subject to legal intervention.
-- James A. Woods (ames!jaw)
------------------------------
Date: Thu, 14 May 87 05:58:02 PDT
From: Ya'akov_Miles%UBC.MAILNET@MIT-Multics.ARPA
Subject: A Fix Please for Backspace and Long Command Lines
In Issue 036 of INFO-IBMPC digest, a brief note by Dana Myers
appeared as a response to my complaint about Backspace
failing to delete long PC-DOS 3.1 command lines. I am a
bit more curious to find out how to "fix" the bug, rather
than be convinced that the bug is a "feature"...
- Ya'akov Miles,
"Applications" Engineer,
TRIUMF,
(An open Facility dedicated to medical
and other applications of Nuclear
Physics, and paid for by the
Government of Canada...)
The University of British Columbia,
4004 Wesbrook Mall,
Vancouver, B.C.,
Canada, V6T 2A3
------------------------------
From: David R. Conrad <davidc@terminus.UMD.EDU>
To: lotto@harvard.HARVARD.EDU
Subject: IBM's PC TCP/IP
Date: Thu, 14 May 87 19:08:25 EDT
>Does this mean that all PC software purchases MUST be tied to a
>specific VM host license?
Nope, we were worried about that ourselves, so we checked into it
and we were told that purchase of the VM host stuff wasn't necessary.
>Also, these are GRADUATED fees. To buy a 50 copy license, you must
>purchase BOTH the 20 and 50 copy options...
Yes, that's true. The price ranges from $120 per copy at quantity 20
to $96.40 per copy at quantity 250 ("One Time Charges").
>>also pay a one-time charge of $350 for the master diskette that you
>>copy with a special program.
>Are these distributed versions then copy protected? If so, can it run
>off of a hard disk? Can you backup the hard disk with the program
>installed?
No, there is no copy protection. Therefore no problems with hard disks.
IBM has a program that will allow you to make N number of licensed
copies where N is the number that you paid for. It's really an
honesty policy and database to keep track of where the copies have gone.
I personally would recommend a hard disk. The full package should
use up quite a bit of disk space (on the order of about 1.5M).
Note that we haven't seen what IBM is going to be distributing, so
our assumptions are based on what we gave them.
>All in all, it is not clear that this is particularly useful in
>anything but a VM environment. The TELNET probably can call any
>telnet server, but what is "ASCII" support, ANSI or just a line
>by line glass terminal? Maybe it emulates an LA-100 :-)
The package is based on the (in)famous MIT code and is somewhat more
useful (with the inclusion of mail and ftp). The telnet program
emulates the H-19 terminal in ASCII mode (a la the original MIT and CMU
versions). If connecting to a VM host through Wiscnet, full screen mode
will be entered emulating a 3278-2.
>I would imagine that the TFTP, SMTP and FTP are full function
>client implementations with no servers. So we can be a "one way"
>workstation. So what does it do with "FROM"?
There is both a TFTP server and TFTP client, as well as TFTP support
from within telnet. SMTP is part of the mail system and performs only
client functions. FTP is also client side only.
If by "FROM" you mean the "from:" line in the mail message or the
SMTP "MAIL FROM" command, then the FROM stuff is filled in by the mail
program 'send' with your POP-2 mailbox.
>Is there ICMP support in the PC implementation?
The package handles ICMP echo requests and includes a version of
ping(8). All other ICMP datagrams are accepted but are not upcalled to
the TCP or UDP layers.
>And what about the Token ring version? I know about rfc 1001, 1002 for
>NetBios on TCP/IP, but is there any rfc for TCP/IP on NetBios? Do
>they conform to the NetBios interface in this implementation? How
>about a NetBios compliant version that writes to the hardware for
>screen and keyboard IO? (I do not know this is the case, I am just
>wondering. Will the license even permit the use of this program on
>clones?)
Token ring and NETBIOS support are two different issues. The package
supports the IBM Token Ring Card via the direct interface of the TOKEREUI
program. The package can run concurrently with other software
that uses the token ring (such as PC LAN). IP packets are wrapped in
RFC990 SNAP and a 802.2 LLC header.
There is no support for NETBIOS on TCP/IP, nor do we run on top of NETBIOS.
>I heard a rumor that TCPINFO@YKTVMV.BitNet might shed some light on
>the situation if the IBM gateways are not filtering mail for them.
I wouldn't think IBM would do that, but I'm not positive on this.
>Gerald Lotto - Harvard Chemistry Dept.
-drc
David R. Conrad The University of Maryland arpa: davidc@umd5.umd.edu
(301) 454-2946 PC/IP Group bitnet: conradd@umdd.bitnet
"Although golf was originally restricted to wealthy, overweight Protestants,
today it's open to anyone who owns hideous clothing." -- Dave Barry
------------------------------
To:tweten@ames-prandtl.ARPA (Dave Tweten)
Subject: How to Detect an EGA?
Date: Thu, 14 May 87 20:42:46 EDT
From: James R. Van Zandt <jrv@mitre-bedford.ARPA>
> ...So who knows the best way for software to discover if the adapter on the
> machine is an EGA?
I've heard that a lot of people merely check whether the ASCII characters
"IBM" appear at address C000:001E. However, IBM recommends this procedure...
static is_ega()
{ /* returns with AX nonzero if EGA is present */
#asm
mov ax,1200h ; set ax for BIOS call (alternate select)
mov bx,0ff10h ; set bl for Return EGA Information
; load bh with invalid info
mov cl,0fh ; load cl with reserved switch setting
int 10h
mov ax,1
cmp cl,0ch ; test reserved switch setting
jl is_2
xor ax,ax
is_2: cmp bh,01h ; check color/mono information range 0-1
jle is_3
xor ax,ax
is_3: cmp bl,03h ; check memory size bits range 0-3
jle is_4
xor ax,ax
is_4:
#endasm
}
- Jim Van Zandt
------------------------------
Date: Fri, 15 May 87 10:09 CDT
From: Weinstein@HI-MULTICS.ARPA
Subject: 720K 3.5" floppy with DOS 3.3
How does one configure a 3.5" floppy on an AT type system using DOS 3.3.
I have tried several device directives in my config.sys, however I have
never hit the correct method.
Can someone suggest the correct approach... I want to make the drive a
720K floppy all the time.. just like the IBM Model 30 drive.
------------------------------
From: tr@thumper.bellcore.com (tom reingold at thumper.bellcore.com)
Date: 15 May 1987 1200-EDT (Friday)
Subject: Redirection in Batch Files
Michael:
Precede the name of your batch file with "command /c" and
you can direct it. For example, type
command /c compile > errors
Tom Reingold
INTERNET: tr@bellcore.bellcore.com
UUCP: ..!decvax!ucbvax!ulysses!bellcore!tr
..ihnp4!bellcore!tr
------------------------------
To: oxy!bagpiper@csvax.caltech.edu (Michael Paul Hunter)
From: Jeff%acorn@oak.lcs.mit.edu
Date: Thu, 14 May 1987 14:25 dst
Subject: Redirection in Batch Files
I don't see why you couldn't do the following:
pass1 %1 > %1.err
pass2 >> %1.err
link %1 >> %1.err
Of course, one could get fancy with a COMPILE.BAT like the following:
if not %1x==x goto got_file
echo Need at least one arg.
goto foo
:got_file
if %2x==x compile %1 con
pass1 %1 > %2
pass2 >> %2
link %1 >> %2
The second depends on the .BAT chaining, allows you to default by
sending error messages to the screen, with a second optional
parameter specifying the file. (One could easily switch this
around, defaulting to other things and forcing you to specify
COMPILE FOO CON
If you wanted to see the messages on the screen. I think that DOS
2.x doesn't support file appending, I know 3.1 does, and suspect
that 3.0 does. (Rumors I've heard.)
-jeff
------------------------------
Date: Fri, 15 May 87 13:10:02 EDT
From: Russell Nelson <bh01%CLUTX.BITNET@wiscvm.wisc.edu>
To: swingler@baylor.bitnet
Subject: Z-248 (8 Mhz 0 wait state) memory upgrade
We have about 900 Z-248s on campus. I have no idea of how many have upgraded
memory. We used to use the Everex Magic 16, but had troubles with it. I
believe that using 100ns chips solved the problem but by the time we
discovered that we had switched to Addionics AT expansion boards. Both boards
give you two serials, one parallel, 640K, and 1.5Meg. You're right, these
boards are hard to find, but we've found something that works for us.
-russ
GEnie: BH01
BITNET:BH01@CLUTX
uucp: decvax!sii!trixie!gould!clutx!bh01
------------------------------
Date: 15 May 87 22:18:00 EST
From: "GERRY SCHMID" <schmidg@esdvax.arpa>
Subject: Problems Repacking DOS Partitioned Split Disks with REFORMAT
I N T E R O F F I C E M E M O R A N D U M
Date: 15-May-1987 22:17
From: Gerry Schmid
Username: SCHMIDG
Dept: PL
Tel No: 617-256-3969
TO: _MAILER! ( _DDN[INFO-IBMPC@C.ISI.EDU] )
TO: _MAILER! ( _DDN[KIRSCH@BRAGGVAX.ARPA] )
CC: _MAILER! ( _DDN[SCHMIDG@ESDVAX.ARPA] )
I'm glad V6.036 mentioned the JOS REFORMAT program for improving
disk performance. I have a copy of the "MS-DOS Version by (CFS)"
with which I am trying to repack a 40Mb Seagate ST451 on my PC's
Limited 8Mhz AT. I also transferred and compiled Versions 1.2 and
1.21 from the Simtel20 but neither version makes it past "Reading
Bootsector, Fixed Disk: cannot compute size". I have not tried
any of the commercial programs, but have been advised by PC's
Limited that some programs may corrupt data on split disks.
With disks larger than DOS currently allows (30Mb), it's
necessary to partition the disk and run split disk system
software during booting. REFORMAT was tried with two of these
split disk programs, BTATDISK.SYS for a 40Mb Miniscribe on a 6Mhz
IBM-AT and SPLIT_1.SYS for my PC's Limited. REFORMAT works on both
partitions of the Miniscribe as designed, but on the PC Limited
Split Disk software it cannot find the File Allocation Tables in
the second partition, i.e., Finding FATs and root directory:
Unable to find FAT on disk.
A person I spoke to at PC's LTD tech support (1-800-624-9896)
suggested that the FAT for the split disk was probably on
cylinder 410 or 411 (The ST451 has 820 cylinders which I split
into two 20Mb disks).
I realize that there are many unknowns here, but I wanted to
respond to the question about problems with REFORMAT.
Any Suggestions?
Gerry Schmid
------------------------------
Date: Sun, 17 May 87 12:30:44 GMT
From: Gregory Hicks COMFLEACTS - Chinhae <hicks@walker-emh.arpa>
Subject: DIAL.SRC unpacks OK
Responding to the editor's request of a few issues back, the source
for DIAL.SRC does indeed unpack itself when used as 'stdin' for a
UN*x shell. However! before running the script, the first six lines
that begin with the '#' character should be removed.
usage is: /bin/sh < dial.src
then you will end up with the files listed in the first few lines.
Of course, a good editor can be used to unpack the files as well.
Gregory Hicks
------------------------------
To: info-ibmpc@c.isi.edu
Subject: Contribution of GRAPH
Date: Sun, 17 May 87 14:04:05 EDT
From: James R. Van Zandt <jrv@mitre-bedford.ARPA>
I would like to contribute an updated version of my GRAPH program to
the lending library. There are four new device drivers:
gega.c EGA in mode 16 (640 * 350 * 16 colors)
gh.c Hercules monochrome graphics
ghp.c Hewlett Packard 7470A plotter
gli.c CIE LIPS-10 ("Laser Image Printing System")
Notes...
I've had trouble with flow control for the HP plotter, and have now
lost access to the one I've been testing software on. If you find
improvements, please send a copy to me.
My production plots on the LIPS-10 are turning out beautifully, but in
tests I encountered an apparent firmware problem. If the density of
lines gets high enough, some line segments are displaced downwards by
exactly 256 dots. The firmware date printed out on the status page is
December '86. The fellow at CIE mentioned a problem with the "strip
buffer", which sounds reasonable. He didn't promise the problem would
be fixed, but presumably they are working on it. I certainly hope so -
it's a great machine. In addition to vector graphics, the LIPS-10
emulates the Epson MX-80, the Diablo 630, and the HP Laserjet. Ours
cost only $2300. Supplies are expensive, but second sources should
appear.
Here's a revised blurb...
GRAPH GRAPH takes pairs of points (two numbers per line)
(subdirectory) from a file or standard input as x- and y- values
bench.c and plots them on the screen, connected by
graph.c straight lines. Axes with convenient values are
scale.c chosen automatically. Labels, lines of different
g31.c styles, widths, and/or colors, or markers of
g32.c different shapes may be specified. There are now
gh.c device drivers for: Zenith Z-100, CGA, EGA,
ghi.c Hercules, HP 7470A and Houstin Instruments DMP-29
ghp.c plotters, and CIE LIPS-10 laser printer. Other
gli.c revisions since 6/8/86 include capacity for 5000
gpc.c data points, and multiple input files.
gz.c Equivalent to the UNIX program GRAPH. G31.C and
font9.c G32.C are a portable implementation of the CORE
gh.c graphics standard. The above are in directory
core.doc <info-ibmpc.graph>.
graph.doc <James R. Van Zandt, jrv@mitre-bedford> 5/17/87
[These files are found in <info-ibmpc.graph> on C.ISI.EDU -wab]
------------------------------
Date: Mon, 18 May 87 13:16:37 CDT
From: "Rich Winkel" <MATHPG1%UMCVMB.BITNET@wiscvm.wisc.edu>
Subject: Hercules & CGA Incompatibility
I believe the second page of the herc's graphics memory occupies 32K starting
at segment B800. This conflicts with the CGA's ram address space, which makes
the two fundamentally incompatible. (unless you can do a hardware mod on the
herc which disables the second graphics page)
Rich Winkel
------------------------------
Date: Mon, 18 May 1987 14:37 CDT
From: <ARCHERB%UMKCVAX1@wiscvm.wisc.edu>
Subject: Software for the Handicapped
The following is a request for information from one of the users in our
program. Please send any responses to the BITNET address that follows
the text of the request. Thanks in advance to all that respond.
I am interested in finding out about what types of hardware and soft-
ware are available to help the handicapped. I am specifically interested in
finding out what could be helpful to a person whose motor functions have
been severely limited as a result of a gunshot wound to the head. This person
suffered a 50% loss of cerebellar tissue as a result of the bullet wound.
Other impairment seems to have taken place as a result of what I
would call "concussive damage" to brain tissue. (I am not a doctor, so my
descriptions may be bad or erroneous.) Right now this person is fed through
a tube. She cannot speak because her throat is paralyzed. She can move arms,
legs, hands, and fingers, but the movements are very gross. Her eyes function,
but they do not always move together, and she has some trouble focusing.
If any PC users know of hardware or software available which could help
such a person communicate and/or control the environment to some extent please
send a message to me. If you know of agencies or specialists who might be
helpful, please send that information also.
Brian McKeever - System Programmer
University of Missouri Kansas City
Computer Science Program
4747 Building Rm. 219
5100 Rockhill Rd.
Kansas City, MO 64110
(816) 276-2368
------------------------------
Date: Thu, 14 May 87 8:32:59 EDT
From: Kenneth Van Camp -FSAC- <kvancamp@ARDEC.ARPA>
Subject: Mailing List for Handicapped
Speech synthesis devices have been discussed at length on the "computers for
the handicapped" mailing list. Try <L-HCAP@NDSUVM1.BITNET> or
<L-HCAP%NDSUVM1.BITNET@WISCVM.ARPA>
--Ken Van Camp <kvancamp@ARDEC.ARPA>
------------------------------
Date: Mon, 18 May 87 14:27:58 CDT
From: "Rich Winkel" <MATHPG1%UMCVMB.BITNET@wiscvm.wisc.edu>
Subject: PROTECT.ASM problem
This program, which protects hard disks against trojan horses, has a problem
in the beginning where it checks DL to determine which drive is being
referenced during an INT 13. The thing to check for is whether the high
bit in DL is on, but instead it checks for DL=2 or DL>4, which is incorrect.
The series of CMP DL,.. and JZ CONTINUE 's at the beginning should be
replaced with something like:
TEST DL,80H
JZ CONTINUE
As it stands, it interferes with my ramdisk, which is set for C:.
Rich Winkel
[I haven't updated PROTECT.ASM, but have added this as a comment to the file.
-wab]
------------------------------
Date: Mon 18 May 87 21:45:52-PDT
From: Brad A. Silverberg <SILVERBERG@CSL.SRI.COM>
Subject: Turbo C
To: info-ibmpc@C.ISI.EDU
For those who may be interested -- Turbo C began shipping
last Thursday. We are filling direct orders first, then dealers
and distributors. It will take a bit of time to catch up on the
30,000+ back orders, but they should all be filled by the end
of the week.
Borland also announced it will be licensing the source code
to the Turbo C Run Time Library. It will cost $295 and
be available around July 15.
Brad Silverberg
Borland International
------------------------------
Date: Tue, 19 May 87 16:53:27 GMT+0100
From: Keith Dale <kdale@bbncc-eur.arpa>
Subject: S & S Wholesalers BEWARE!
To: info-micro@brl.arpa
Cc: info-ibmpc@c.isi.edu, info-atari16@score.stanford.edu,
info-amiga@red.rutgers.edu, bboard@bbn.com, bbn-eur@bbncc-eur.arpa,
info-mac@sumex-aim.stanford.edu
The saga (nightmare) concludes.....sorta.
Some of you might remember my flame-o-gram of a couple of months ago
recounting my problems with getting a PC system (brand irrelevant)
from S & S Wholesalers out of Miami Beach, FL. In short, I ordered a
new system, same day mailing (as advertised), with the latest OS
release. I got a refurbished CPU, old software, and a broken monitor
- all mailed 2 weeks after the order, sent by 3rd class U.S. Mail
that took a month to get to me. (I won't mention that I was
overcharged, but was charged promptly.) I returned everything to them
for replacement - two months later I received a CPU whose box had
been opened with no software, followed a week later (today) by the
monitor (packed in the same box in which I sent the broken one back,
hole and all) and the software. I haven't tested the monitor yet, so
that's still pending, but the kicker is that they sent me the same
box of old software!!
I just got off the phone with them and asked about the software.
"Why did I get the same software back??" I asked. "Because it was
the only one we had," they replied. "What can we do to get me the
software that I ordered and paid for?" They replied, "We suggest you
go to another company."
Really. That last quote is verbatim. So, I suggest to you that if
you have any microcomputer or peripheral needs, please do what they
suggest - "go to another company".
Keith M. Dale
<kdale@bbncc-eur.arpa>
BBN Communications Corporation
------------------------------
Date: Mon, 18 May 87 11:16 PST
From: <IVAN%TRIUMFCL.BITNET@wiscvm.wisc.edu>
Subject: Bug in Microsoft FORTRAN V4.0 Compiler
To: info-ibmpc@usc-isib.ARPA
X-Original-To: kost,bd,info-ibmpc@usc-isib.ARPA, IVAN
I ran across a bug in Microsoft FORTRAN V4.0 while trying to compile
a benchmark programme published in the Atari Digest. The following code
excerpt demonstrates the bug (which also appears if the outer DO loop is
commented out):
real*4 a,b,c,d,e,f,g,h,p,q,r,s,t,u,v,w,x,y,z
integer*4 i,j,k,l,m
p = 1.0
do 20 i = 1, 20
q = float(i)
do 20 j = 1, 20
r = float(j)
do 20 k = 1, 20
s = float(k)
do 20 m = 1, 20
t = float(m)
v = 1./t
w = s * v
x = r + w
y = q - x
a = y * v
b = a * t
c = b - w
d = c / q
f = d - r
g = f * s
20 p = p + g
end
My compile command was:
fl /c /FPi /AL bug.for
and the compiler error message is:
bug.for(23) : fatal error F1001: Internal Compiler Error
(compiler file '@(#)regMD.c:1.74', line 2156)
Contact Microsoft Technical Support
I presume that the compiler got lost somewhere while trying to
optimize the loops.
------------------------------
Date: Tue 19 May 87 09:52:54-PDT
From: kb5mu@pnet01.CTS.COM (Paul Williamson)
Subject: fix to DU
Organization: People-Net [pnet01], El Cajon, CA
In the program DU by Tom Vijlbrief, recently distributed on the net,
the function comp_cluster tries to read the partition boot record to
get the number of sectors in a cluster. Unfortunately, the method it
uses to find the boot record is incorrect. It simply searches the
first cylinder for a sector that looks like a boot sector, in that it
starts with the byte 0xEB. This is not sufficient in the case of a
hard disk. On mine, it finds an old boot sector left over from a
previous formatting with DOS 2.0, instead of the current DOS 3.1 boot
sector. If you're going to look for the boot sector, the Master Boot
Record of the hard disk must be consulted to find out which sector
contains the boot record for the correct partition.
There is an easier and better way. DOS function 1Ch returns the
needed information, plus the sector size so 512 needn't be hard-coded
into the program. The following is a replacement for the function
comp_cluster() in the program du.c. In addition, the routines
readsect() and max_head() and the associated #defines (BLOCK_SIZE, DISK,
READ_SECT, RESET_DISK, FLOP_RETRIES, GET_PARAM) can be deleted, along
with the declaration for sector[BLOCK_SIZE].
This modification may be distributed freely.
int comp_cluster(drive_nr)
int drive_nr;
{
int nr_sectors;
int block_size;
union REGS regs;
struct SREGS sregs;
segread(&sregs); /* set up input segment registers */
regs.h.ah = 0x1C; /* Get allocation info for drive */
regs.h.dl = drive_nr + 1; /* zero is used for default here */
intdosx(®s, ®s, &sregs); /* get info */
/* intdosx is needed to restore DS after the call. */
nr_sectors = regs.h.al; /* this is what we wanted */
block_size = regs.x.cx;
return(nr_sectors * block_size);
}
Paul Williamson
... !sdcsvax!macomw!williams
or ... !sdcsvax!crash!pnet01!kb5mu
[I don't have time to check out the changes. This note has been prepended
to the du.c source code. -wab]
------------------------------
Date: Thu, 14 May 87 08:32:16 EDT
From: jpp@ORNL-MSR.ARPA (J. L. Patton)
Subject: Fastcard
Does anyone know anything about the Fastcard multifunction/expanded memory
card marketed by Peripheral Marketing, Inc. Any and all comments welcome.
Thanks in advance.
------------------------------
Date: Thu, 14 May 87 15:26:10 EDT
From: Photios_Ioannou@um.cc.umich.edu
Subject: Defeating Model 339 ROM POST
I have an IBM AT model 339 (512 kB RAM, 30 MB hard disk, enhanced
keyboard, etc.). This machine runs (officially) at 8MHz. IBM has
made sure the user does not speed it up (by changing the clock crystal,
a $3 operation) by having the POST part of the BIOS do a speed check:
If the BIOS POST detects that the current speed is more that about
8.45 MHz the machine beeps twice and halts. It does not even go into the
memory check portion of the POST (the part where the memory check is
shown at the upper left portion of the screen). I currently use
a variable frequency synthesizer (called XCLEX) which I bought from
Ariel Corp. and which allows me to change the speed by flipping
a switch after the end of the POST (from 6MHz to 10MHz). XCLEX is
supposed to do the switch from low to high speed automatically by
detecting Power-On and Power-Off, and CTRL-ALT-DEL softboot. However,
its reliability in detecting power-on-off is about 50% and it never
detects soft reboots (which always ends in beeps from the BIOS and a
machine hang). I would like to put an end to this problem by modifying
the necessary code in the BIOS (something tells me that changing a
couple of instructions would be enough) to disable either the speed-check
itself, or the subroutine that follows it if the speed is "illegal".
I have looked at the BIOS by capturing it into a disk file, but were
not able to figure it out.
Our electronic technician can make me new chips from the patched BIOS
provided we find the patch... I would appreciate any help in locating
the culprit code and patching it (using debug, the ULTRA utilities,
or the Norton Utilities). If you have the answer please send it directly
to me:
Prof. Photios G. Ioannou
Civil Engineering Department
2354 G.G. Brown Lab
The University of Michigan
Ann Arbor, MI 48109-2125
Tel: (313)764-3369
PS. I have the AT technical reference manual including the BIOS listing.
It describes the original BIOS in the first AT's that did not have this
problem. As far as I know IBM never issued the listing of the BIOS for
the newer AT's that includes the speed checking code. I have spend a lot
of time on this, so any help is greatly appreciated.
------------------------------
Date: Fri, 15 May 1987 07:42 PDT
From: PAAAAA7%CALSTATE.BITNET@wiscvm.wisc.edu
Subject: Opening Floppy Drive Door Crashes DOS 3.2
To: info-ibmpc@c.isi.edu
Dual drive IBM PC's with Teac half height drives. If you open the
drive door and attempt to access the drive with DOS 3.2, you get
a "General failure reading drive" message, then the drive is
logically removed from the system! The only way to get it back
is to reboot. DOS 2.1's error message is "Drive not ready" which
is correct. Closing the door with 2.1 and retrying corrects the
problem. So is the error with DOS 3.2 or our systems?
Rich
<PAAAAA&@CALSTATE.EDU>
------------------------------
Date: Fri, 15 May 87 12:40:34 EDT
From: "Don N. Kleinmuntz" <DKLEIN%SLOAN.BITNET@wiscvm.wisc.edu>
Subject: Query about AT&T PC 6310
I would appreciate hearing from anyone who has been using the AT&T PC
6310 (8 MHz, 80286-based) regarding (a) general evaluation of the
machine, and (b) IBM PC AT compatibility. The literature I have seen
indicates a high degree of compatibility. I am particularly
interested in your experience with the graphics displays. The
literature mentions "high resolution" graphics (640 by 400 monochrome
and, as an option, color), but does not specifically state that the
display is EGA-compatible. Is it? If not, what are your experiences
with finding supported software?
Thanks in advance for the information.
Don Kleinmuntz
Sloan School, MIT
DKLEIN@SLOAN.BITNET
or
DKLEIN@SLOAN.MIT.EDU
------------------------------
Date: Fri, 15 May 87 11:20:03 pdt
From: well!rogue@lll-lcc.arpa (L. Brett Glass)
Subject: IBM Voice Communications Option Sought
The IBM Voice Communications Option is no longer on the market due to
poor industry response, but I may be able to use one for a special
project that requires it. If you have one you might be willing to sell,
please send electronic mail directly to the address below. I do not
always get the IBM-PC digest, so this is the best way to reach me.
Sincerely,
L. Brett Glass
Usenet: ihnp4!lll-lcc!well!rogue
ptsfa!well!rogue
ARPANET: well!rogue@lll-lcc.arpa
(MM may require quotes, thus: "well!rogue"@lll-lcc.arpa)
------------------------------
Date: 16 May 1987 16:21-EDT
Sender: PACJ77@A.ISI.EDU
Subject: WANG Plotter
From: PACJ77@A.ISI.EDU
I too have a WANG 2272-2 pen plotter that I am trying to drive
from an IBM clone. (Ref V6#23) I tried to reach Stu Mitchell
from the ARPA NET but have had no luck. I would appreciate any
help in this area (plotter) or a way to contact Stu.
Thanks and Aloha, Ron Uyehara
------------------------------
Date: Sat, 16 May 87 13:47:30 EDT
From: "James H. Coombs" <JAZBO%BROWNVM.BITNET@wiscvm.wisc.edu>
Subject: Sort Programs
Does anyone have experience with sort programs for the PC? I have 4
megabyte files and expect to have 12 megabyte files to sort. I have
tried the shareware QSort, but it takes longer to sort these files than
it takes me to generate them. I have seen advertisements for Opt-Tech
sort, and I know that Microsoft has a sort program. I also vaguely
recall a review in PC Tech or some such.
Any information appreciated. Mail me and I will summarize for the
digest. Thanks. --Jim
------------------------------
Date: 17 May 87 0:30-EST
From: R.Rasulis,Jr. <smggyuc%BOSTONU.bitnet@WISCVM.arpa>
Subject: Text Files: DEC <-> PC
I'm sure someone may have made a similar query, but I've run into
this scenario for the first time:
A colleague uses a DEC Rainbow and DOS. I'm using a PC files. I
tried to read a disk from his machine that had only text files. But I
couldn't even get a directory listing. Is there some trick to read
DEC's MS-DOS Text files using a PC-DOS machine and vice-versa?
Please respond directly, if enough responses, blah-blah-blah.
*eom :- R.Rasulis, Jr. -:
------------------------------
Date: 17 May 1987 2049-PDT (Sunday)
From: melmoy@nprdc.arpa
Subject: 5 Mhz Math Coprocessor on Turbo xt
Will someone set me straight on the question of whether the
8087-3 (5 MHz) coprocessor can be safely used on a turbo xt
motherboard? Or is it absolutely necessary to use the 8087-2 (8
MHz) because of circuitry requirements? The numbering schema on
math coprocessors doesn't fit my intuitions.
Mel Moy
melmoy@nprdc.arpa
------------------------------
Date: Sun, 17 May 87 10:16:27 cdt
From: Esmail Bonakdarian <bonak%cs.uiowa.edu@RELAY.CS.NET>
Subject: Telecommunication Outside the U.S.
Hello, does anybody have any experience with the telecommunication
situation outside the United States? More specifically, I have a
pc and an u.s. robotics modem plus copies of kermit and procomm. I
am planning to move to the far east (Taiwan, perhaps mainland China)
soon and am wondering if i'll need to modify hardware (new modem?)
and/or software (how widespread are kermit and procomm type "standards")?
Any information (even about other countries other than the PRC and ROC
are also most welcome).
Many thanks,
Esmail Bonakdarian
------------------------------
Date: 14-May-1987 1647
From: decwrl!parity.dec.com!ornstein@ucbvax.Berkeley.EDU
Subject: Microsoft Cobol is 1974 Standard
Can someone please explain why MicroSoft brought out a new
$700 release for Cobol based on a 1974 standard.
Am I the only one interested in the Cobol-81 or 85
standards. All the structured programming enhancements
went into the language after 1974.
- Ian -
------------------------------
From: Cliff Joslyn <rlgvax!cliff@seismo.CSS.GOV>
Date: 16 May 87 21:56:32 GMT
Subject: Dbase, Lotus information
Organization: Computer Consoles Inc, Reston VA
I am looking for the best of all possible published manuals on Dbase
II/III/III+ and/or Lotus 1-2-3 for the IBM PC/XT/AT. I would like it to
be addressed to the experienced programmer, and include complete
language specifications, data dictionary, file formats, available
support libraries, error codes, startup file protocols, advanced
programming examples, common programming and optimization practices,
known bugs, hardware compatibilities, etc. etc. etc.
A tall order? Help would be appreciated.
--
O----------------------------------------------------------------------->
| Cliff Joslyn, Computer Consoles Inc., Reston, Virgnia, but my opinions.
| UUCP: ..!seismo!rlgvax!cliff
V All the world is biscuit shaped
------------------------------
Date: 1987 May 17 22:09 EDT
From: (Bob Babcock) PEPRBV@CFAAMP.BITNET
Subject: Pop-Up Dictionary
A friend of mine is looking for a pop-up dictionary for use by a
foreign language translator. Do such programs exist? (I didn't ask
what languages were needed, but I think there were several.)
------------------------------
Date: Mon, 18 May 87 14:43 CDT
From: <ARCHERB%UMKCVAX1@wiscvm.wisc.edu>
Subject: 286 Accelerator boards for XT
We are thinking about obtaining an 80286 accelerator board for our
XT ( with a 80287 co-processor ). Anyone have any that they are either
very happy or very unhappy with? I know nothing about the pros and cons of
the various beasties.
Thanks,
Barry Archer
ARCHERB@UMKCVAX1.BITNET
------------------------------
Date: Mon 18 May 87 16:41:32-PDT
From: BUSSARD@EDWARDS-2060.ARPA
Subject: Int 32 and 52
Can some one give me some information on undocumented interrupts on MSDOS
Versions 3.1 and 3.2. Most interested in 32 and 52.
Buzz
Bussard@Edwards-2060
------------------------------
Date: Mon, 18 May 87 21:38 EDT
From: David M Chizmadia <Chizmadia@DOCKMASTER.ARPA>
Subject: Pinouts for the M24 computer.
My father-in-law just picked up an Olivetti M24 (aka ATT 6300) computer.
He didn't get a monitor for 2 reasons: the had run out and he already
has an RGB monitor.
His problem:
He has gotten the pinouts for the monitor connector, but he doesn't
know what anything beyond the HSYNC and VSYNC line really does.
Specifically, he knows that the pins labelled D1-D3 register activity
when he types anything, but when he hooks them up to his monitor, all he
gets is a pattern of dashes.
If anyone can shed any light on the meaning of the various signals,
please respond directly to me. My father-in-law fixes TVs for a living
so get as technical as necessary, I'm just passing it on. If enough
people are interested, I'll post back what I find to this group.
Thanks for your help,
Dave Chizmmadia
Chizmadia @ DOCKMASTER.ARPA
------------------------------
Date: Mon, 18 May 87 11:25:38 cdt
From: Esmail Bonakdarian <bonak%cs.uiowa.edu@RELAY.CS.NET>
Subject: shipping a pc overseas
Hello again, does anybody out there have experience with, or suggestions
about shipping a pc (and associated peripherals) overseas in a reasonable
safe and "cheap" way?
I'd appreciate any info on this.
Thanks.
------------------------------
Date: Tue, 19 May 1987 09:20 PDT
From: JAJZ801%CALSTATE.BITNET@wiscvm.wisc.edu
Subject: EMM and DMA
I have read many of the recent articles on EMM memory and
boards (PC Tech Journal, May 1987, PC magazine board/emulator
reviews) but have not seen the following issue discussed. Can
someone please expand (sorry) on the subject:
1. Is there a meaningful (bus accessible) physical address
for EMS/EEMS pages (the physical memory) when they are
not mapped into a page window. I suspect this may not
be true for PC-type boards since the above 1-megabyte
set of addresses required for the up to 8 megabytes of
EMS address space would be beyond the addressability of
8088-class processors. What about AT-type EMS boards
since they also sometimes can be split to fill extended
memory ?
2. If there are bus-addressable memory locations for either
the PC or AT implementations, are they accessible when
the pages are not mapped to a window ? For instance, if
physical memory address were determinable (another problem
since no calls relate to it), could DMA operations be done
on pages while not mapped, or would the EMS board refuse
to respond to the address selection ?
I suspect that the answers to these questions are negative
but would like to know for sure.
Jeff Sicherman
JAJZ801@CALSTATE.BITNET
------------------------------
Date: Tue, 19 May 87 12:04:09 CST
From: "Dick Smith" <FAC0069%UWF.BITNET@wiscvm.wisc.edu>
Subject: More on the ANSI question
If I understand the whole business of the ANSI drivers, these constitute
an agreed-upon standard of non-ASCII codes, which when received by a terminal,
the same non-character sequences (like CBT, DL, IL, etc) produce the same
results on the actual piece of hardware. The NAPLPS standard for graphics-
oriented communication does the same thing, except here the code set includes
a bunch of graphics primitives, which are carried out at the receiving end.
Do I have the picture right?
Here are my two questions for the day: why is the ANSI.SYS driver so slow?
Seems that a principal purpose of using these codes is to effect a desired
result in less time, not more. I used the demo in Norton's NU program to
compare speeds.
And second: I couldn't see why, in Leonard Vanek's posting of codes, the
subject of the VT-100 code set got into it. Which came first, X3.64 or
the VT-100 device? Did one serve as the basis for the other?
==>Dick Smith <FAC0069@UWF.BITNET>
------------------------------
End of Info-IBMPC Digest
************************
-------