INFO-IBMPC@USC-ISIB (04/10/83)
From: Dick Gillmann <INFO-IBMPC@USC-ISIB>
Info-IBMPC Digest Sunday, 10 April 1983 Volume 2 : Issue 22
Today's Topics:
DOS 2.0, 2.2 and 3.0 (5 msgs)
COMPAQ
Lisp Interpreter
----------------------------------------------------------------------
Date: 7 Apr 83 10:08:07-PST (Thu)
To: info-ibmpc at Usc-Isib
From: decvax!yale-com!glaser (Robert D. Glaser) at Ucb-Vax
Subject: DOS 2.0, 2.2?, and 3.0
About DOS 2.2: I think your source is brain damaged. IBM doesn't
usually do stupid things like releasing a new version of a product
before the previous version is out. The only possible explanation
could be that DOS 2.0 has really bad bugs in it, but I haven't heard
of any.
I've used 2.0 a bit (I have it but my documentation hasn't arrived
yet) and I agree that it's really a major improvement. Piping,
filters, and tree structured directories are implemented 99% like
unix, with command names like mkdir, rmdir and cd, and supplied
filters named more, sort, and find (= grep -- a sensible name change).
Not having multitasking hurts, but sources at Microsoft say that DOS
3.0, which will be out who knows when, does. Evidently, IBM didn't
let Microsoft expand DOS 2.0 as much as it might have, so the choice
was either multitasking or everything else. I think they chose right.
As it is, 2.0 is 24k, double the size of 1.1. It looks like the 64K PC
might go the way of the cassette interface.
There are, of course, short-term (hopefully) compatibility problems
with 2.0. All 3rd party hard disks and special peripherals, and much
3rd party software, is likely to need fixes for 2.0. This won't be as
bad as it might seem, because 2.0 now supports installable device
drivers. At boot time, 2.0 looks at a file called config.sys for info
on these devices, which could be either hardware (e.g., external hard
disks) or, I think, software (key handlers, terminal emulators).
It'll take a little while for manufacturers to supply these drivers
(most were caught with their pants down), but their existence will
help each compatibility problems with future versions of DOS.
>From personal experience, Lotus 1-2-3 doesn't work under 2.0, but
Visicalc does. My 20 Meg Tallgrass hard disk ( a very good piece of
hardware, incidentally) is hopeless. Computer Innovation's C Compiler
(another terrific product) sort of works -- it's fine if your code is
o.k., but it takes a space shot reading error messages from disk.
For a good summary of 2.0, read the latest (April) Softalk/PC. In
sum, IBM seems to have balanced well the need to not disrupt too much
the installed user base while at the same time moving forward. The
Microsoft/IBM combo is indeed a good one, and is genuinely moving the
business world (or at least the micro business world) towards Unix.
Rob Glaser
(decvax!yale-comix!glaser)
glaser@Yale
------------------------------
Date: 8 Apr 1983 8:22:45 EST (Friday)
From: Edward Haines <haines@BBN-UNIX>
Subject: More on DOS 2.0
To: Info-IBMPC @ isib
Cc: schoff@BBN-UNIX
Here are some more comments on the new DOS 2.0:
There is now an Assemble command in Debug, a handy addition.
You can reassign any keyboard key to generate any string of
characters, similar in function I believe to the new packages selling
for around $75.
There is a thick update for the Basic 1.1 manual included. (Not so
nice for those of us with the old Basic 1.0 manual.)
64K is not enough memory to use 2.0. Not only is DOS bigger but so
are BASIC and BASICA. Also some of the advanced functions (such as
keyboard reassignment) require additional memory. Here is a table of
memory available on my 64K (65536) system (I have ordered a memory
expansion board):
DOS BASIC BASICA
Dos 1.0 (obsolete) 53392 39498 34049
Dos 1.1 53136 38907 33402
Dos 2.0 plain 40960 24050 14001
Dos 2.0, PRINT, ansi.sys 36272 19362 9313
I think the new device driver support is a big winner. It should make
it much easier for vendors to add new functions. I assume the vendors'
software would not need to be version dependent (after DOS 2.0) and
should not clash with other vendors' modifications to the system. But
it looks like all the present mods (e.g. JFORMAT) become obsolete.
You cannot mix system functions between 1.1 and 2.0; e.g., running a
1.1 CHKDSK on a 2.0 system hangs the disk drive.
Pipes are slow compared to UNIX systems since the intermediate file is
written to disk. Note that this requires write access to disk in
order to use pipes. Also the MORE command does not take a file name
as an argument; to use to type out a file you need to put the "<" in
front of the file name to redirect the input.
Question: Would it be possible to write a device driver to map the IBM
Graphics printer codes onto another printer such as the NEC 8023 so
those of us with non-standard printers could take advantage of
software written for the IBM printer? (I am thinking of the Graphics
Screen-dump facility in Dos and graphics applications such as 1-2-3.)
Ted Haines
------------------------------
Date: 8 April 1983 11:41 +0500
From: schoff @ DDN1
To: info-ibmpc @ isib
Subject: Winchesters, DOS 2.0 & 3.0
Date: April 8, 1983
cc: haines @ bbnu
Would anyone like to speculate on Winchesters and DOS?
These are my thoughts and questions:
1. With IBM offering a 10M Winchester, I think that in order
for the other manufacturers to remain competitive that
they will have to lower their prices. Another effect
on this would be that their a lot of Winchesters being
made and this usually leads to lower prices also. I
have seen Davong's 5M going at a mail-order house for
$1400. Does anyone think that we will see Winchester,
controller, cabling, and device driver for $1000 at
the end of this year?
2. With DOS 2.0 it would seem that Microsoft is leading us
down the path to UNIX and Winchesters, certainly some
help is gained through the new filing system with
floppies but the raison d'etre is IBM's Winchester.
3. Does anyone see a DOS 3.0 with some sort of concurrent
tasks? I know that Microsoft has a slow version of XENIX
running on the PC. What would make sense for the minimum
size of Winchester for this XENIX? 10M? If we do have
this multitasking XENIX there will probably be some swapping,
will the standard stepper-motor Winchester be enough even
for a single user?
schoff at bbnu
------------------------------
Date: 9 Apr 83 0:22:05-PST (Sat)
To: info-ibmpc at Usc-Isib
From: menlo70!sri-unix!billw at Ucb-Vax
Subject: Re: DOS 2.2
Well, there was a DOS 1.85 that consisted mostly of a new COMMAND.COM
file, that had something like csh's history mechanism. Maybe DOS 2.2
includes this...
BillW
------------------------------
Date: 6 Apr 83 18:37:48-PST (Wed)
To: info-ibmpc at Usc-Isib
From: teklabs!ogcvax!metheus!cdi!caf at Ucb-Vax
Subject: DOS 2.0
Several announcements have indicated that DOS 2.0 is compatible with
DOS 1.1 (proper superset) and programs written for DOS 1.1 will run on
2.0. (Bill Gates: Yes, absolutely! -PC World Vol 1 #1 p. 52)
In at least one instance, this is not true! In DOS 1.1, function 1Bh
returns a pointer to the current disc's file allocation table along
with other some information necessary to calculate the free blocks
remaining. (The DOS 1.1 documentation on this function was incorrect,
indicating two different items of information were returned in the
same register - See Fig. 1). In DOS 2.0, the pointer only points to
the first byte of that table - the rest has disappeared.
After spending an excessive amount of time debugging and printing
listings (the DOS debugger does not accept program symbols) I finally
got the function to work properly, only to find out that DOS 2.0 had a
Flag Day for this function.
Some other flames about DOS 2.0:
The disk buffers do not support read-ahead or write-behind, even tho
the IBM hardware is designed to allow this.
DOS 2.0 steps the floppies very noisily and seems to cause a
disproportionate number of bad sectors. It is necessary to turn
verify ON in an autoexec.bat file. However, once this is done, the
date/time setting must be done in that file and execution of batch
files is still slow. Sorry, you can't put those commands in the
config file!
Even without disk read errors, DOS 2.0 has an unwelcome habit of
corrupting the file system by losing free blocks. It has become
necessary to run the disk checker program before any serious work to
check on the integrity of the file system.
The ANSI terminal emulation is minimal and lacks any editing functions
(insert/delete). It is not much better than an ADM-3a simulation. In
addition, some ANSI editing command puts the emulator into some
strange 40 column color mode.
The edlin editor remains a classic of cruftiness. It still crashes on
files without carriage returns. In the same article Bill Gates said:
"There's really a lot of dirty software on the market now; we'll have
to educate the developers about how to write better software."
Judging by DOS 2.0, edlin, and Microsoft Pascal, it would appear that
Microsoft will have to look outside their organization for suitable
teachers.
By 2.12 things should be settled down, but in the meantime, Long live
Coherent, Marc and Unix (TM).
Chuck Forsberg
------------------------------
Date: 8 Apr 83 7:46:44-PST (Fri)
To: info-ibmpc @ Usc-Isib
From: harpo!seismo!presby!burdvax!psuvax!psuvm%csx @ Ucb-Vax
Subject: COMPAQ
I answered a general query about the PC-compatible portable COMPAQ
recently with a few positive remarks. Yesterday, my friend said that
one of my BASIC programs doesn't work on his COMPAQ -- it hangs the
system. I'm quite certain it is my routine to switch from the
monochrome display to the color, which involves POKE-ing a bit in
location &H410.
So, if you are going to buy a COMPAQ, make certain any software that
uses the color and monochrome on the PC works as well on the COMPAQ.
Perhaps DOS 2.0 eliminates the problem -- at least at the DOS level
there is a command to switch monitors; is it still done with a poke
from BASIC?
I assume you all saw the COMPAQ review in the latest PC World.
------------------------------
Date: Sun 10 Apr 83 07:56:54-PST
From: Ginger Edighoffer <CSD.HARKNESS@SU-SCORE.ARPA>
Subject: Lisp Interpreter
To: info-ibmpc@USC-ISIB.ARPA
I was frustrated by the lack of a lisp interpreter, so I have written
one of my own. It is written in C for use under the Quinix operating
system put out by Quantum. The interpreter can handle floating point,
but that means you need the 8087 co-processor to run it as is, even if
you run Quinix. I have no desire to spend the time needed to re-write
to convert to run under DOS or without the floating point. Since I
doubt there is much market for the interpreter as it stands, I am
willing to give away interpreter plus sources to any who are
interested.
As it now stands, the interpreter really should be run on at least a
256k memory system. It could be run on a 192k system but you would
really feel the space crunch. It uses about 50k for code, 10 to 64 k
(the more the better) for cons elements and 40 to 64k (again, the
more, the better) for stack, global variables, strings, etc. I have
documentation on what standard functions are provided and I am working
on adding better comments to the source.
Ginger Edighoffer
------------------------------
End of Info-IBMPC Digest
******************************
-------