BYTE Article of Gary Kildall
French Luser
----------
- "CP/M: A Family of 8- and 16-Bit Operating Systems"
Gary Kildall
BYTE, June 1981, p.216
(Retyped by Emmanuel ROCHE.)
This article is about microprocessors and CP/M: where they came
from, what they are, and what they are going to be. Where they
came from is history, what they are today is fact, and what they
will become is, like any projection of technology, pure "science
fiction" speculation. CP/M is an operating system developed for
microcomputers. But as microprocessors changed, CP/M and its
related programming tools evolved into a family of portable
operating systems, languages, and applications packages.
The value of computer resources has changed dramatically with
the introduction of microprocessors. Three major events have
precipitated a revolution in computing: hand-threaded core
memory has been replaced by mass-produced semiconductor memory;
microprocessors have become plentiful; and IBM decided that the
punched card is obsolete. Low-cost memory and processors have
reduced the cost of computer systems to a few hundred dollars,
but IBM's specification of the floppy disk standard has made the
small computer system useful.
In the early days of the Intel 8080 microprocessor, a small
company called Shugart Associates was taking shape up the street
from Intel. Shugart Associates, along with a number of other
companies, viewed the floppy disk as more than a punched card
replacement; at that time, the primary low-cost storage medium
was paper tape (used in applications ranging from program
development to word processing). At a cost of $5, a floppy disk
held as much data as two hundred feet of paper tape, and a disk
drive retailed for only $500 -- an unbeatable combination.
Memory, processor, and floppy-disk technology improved, and by
the mid-1970's, a floppy-based computer could be purchased for
about one quarter of a programmer's annual salary. Quite simply,
it was no longer necessary to share computer resources.
Since that time, microprocessors have been applied to a variety
of computing needs beyond replacement of low-end minicomputers.
Due to applications such as machine-tool movement and sensing,
data acquisition, and communications, current interest lies in
real-time control. In a real-time operating system, process
management can be separated from the I/O (input/output) system
(which is not required in many applications). Real-time
facilities allow the execution of interactive processes
according to priority, and their addition or deletion in a
simple fashion. This results in a custom operating system
designed to solve a particular problem. In contrast to
timesharing, real-time operating systems have minimal "interrupt
windows" in which external interrupts are disabled. Real-time
operating systems such as the Intel RMX and National Starplex
packages provide this level of support.
The emerging interest in local networks poses a new challenge to
designers of operating systems. Recently, Intel, DEC (Digital
Equipment Corporation), and Xerox formed an alliance to promote
Ethernet, a packet-switching network intended to provide point-
to-point data transfer in an office environment. (In a packet-
switching network, data from several slow-speed sources, such as
user terminals, is collected over local lines by a single
network node, which then periodically transmits the data to its
destination at a much higher speed, in groups called packets.)
In terms of evolution and potential, Ethernet is today what
floppy disks were a decade ago. This inexpensive office network
performs such tasks as the transfer of a form letter from data
storage at one location to a memory typewriter in another part
of the building. When modifications are completed, the letter is
typed locally or sent to a laser (or other) printer that is a
shared network resource.
Most timesharing systems handle a network through simple file
transfers between the machines (nodes) in the net, but real
refinements occur when the operating system itself is
distributed among the nodes. File access is provided by one
server node, while a computing function is performed by another.
To the user, a requester node appears as a powerful computing
facility, even though it may consist of only a local
microprocessor, a console, and a limited amount of memory.
What refinements have been made to operating systems? Our models
have been simplified; we understand primitive operations
required for reliable process synchronization in real-time
systems, and the human-oriented interface in interactive
subsystems has been improved. We will, no doubt, continue to
refine our models for timesharing and real-time operating
systems, but the most exciting new operating system technology
will develop around emerging network hardware.
The Emergence of Software as a Problem-Solving Tool
---------------------------------------------------
Microprocessors are a natural consequence of our technology. I
recently visited the British Science Museum, where two
particularly interesting historical developments were on
display. The first exhibit chronicled the development of the
finely machined iron and brass steam engines, complete with
magnificent gauges, gears, whistles, and valves, that founded
the Industrial Revolution.
The second exhibit displayed progress in computing, beginning
with Charles Babbage's inventions of the early 1800s. What did
these exhibits have in common? They showed machines built with
the same technology: Babbage's analytic engine might easily be
mistaken for a small steam engine!
I followed the sequence of displays, from Babbage's difference
and analytic engines to great brass calculators and early punch
cards, past relay and vacuum tube processors to unit record
equipment, then to transistor and random logic computers and
semiconductors and, finally, to a single Intel 8080
microprocessor.
Examined in this way, the technological momentum was obvious.
Microprocessors are a direct result of our pattern of refinement
through engineering. Just as a Boeing 727 is a refined version
of the original Wright Brothers' invention, the microprocessor
is a consequence of "fine tuning" by scientists and engineers
who strive to understand, simplify, and add function to
mankind's tools. There were several conspicuous spaces waiting
to be filled following the 8080 display.
In public television's "Connections" series, James Burke claimed
that we are a society filled with machines that do everything:
sew materials for our clothes, carry us from coast to coast, and
print millions of news papers daily. But the most important
machines in our society do absolutely nothing by themselves.
These multifunctional devices provide a variety of services
depending upon our needs, and herein lies the essential
advantage: in the past, we identified a need and built a machine
to satisfy that need; today, technology provides us with a
single machine that we can instruct, through a program, to solve
almost any problem. Where are the "Thomas Edisons" who used to
build machines? Most are now inventing programs.
The evolution of our electronics industry typifies refinement
through engineering. Beginning with electrical and electronic
switches, we began manufacturing general-purpose function chips:
put a value x on the input pins, define the function f by
setting voltage levels on a second set of pins, and the result,
F(x), magically appears on the output pins. Many examples of
such integrated circuits exist, ranging from three state logic
gates to arithmetic/logic units.
With the introduction of microprocessors, the function f may be
defined through instructions in a read-only memory allowing, in
principle, the implementation of any function using a single
device. A design that once required connecting resistors,
capacitors, and logic gates has developed into a program that
instructs a multipurpose machine to perform the same function.
Controlling a stoplight and balancing a checkbook are now
equivalent problems: both require the invention of a program.
Refinement through engineering: does this not also apply to
software? To properly frame the answer, remember that the
primary purpose of a computer is to be useful. Therefore, the
application program is really the only important result of a
software engineering activity. Our primary goal in refining
software tools is to provide the means for rapid and accurate
generation of simple, understandable, and effective application
programs. We do this through three levels of software support:
system languages, operating systems, and application languages.
These tools form an inverted pyramid underlying application
software.
Application Languages
---------------------
Application languages form the top level of support for
application programming. How does this level of language differ
from other language levels? First and foremost, an application
language contains the operations and data types suitable for
expressing programs in a particular problem environment. FORTRAN
(FORmula TRANslation), for example, was designed in the late
1950s for scientific applications; FORTRAN programs, therefore,
consist primarily of algebraic expressions operating upon binary
floating-point numbers expressed in scientific notation.
However, FORTRAN contains only primitive file-access facilities
and no decimal arithmetic, making it unsuitable for commercial
data processing. COBOL (COmmon Business-Oriented Language) has
the commercial facilities, but it excludes scientific features
such as a complete transcendental-function library.
In contrast to system languages that run on a given machine,
these application languages would ideally contain no machine-
dependent features. An application language is either poorly
designed or ill-suited for a particular problem if the
programmer is forced to use extra-lingual constructs to access
lower-level functions of the operating system or machine. The
language must be a standard, without the necessity for various
locally-defined language extensions. An extended standard
language is of limited value, since the extensions are unlikely
to exist in other implementations.
The evolution of PL/1 (Programming Language/One) provides a good
example of refinement in application languages. PL/1 is not a
new invention; rather, it was defined by a committee of IBM
users in 1960 as a combination of ALGOL (ALGOrithmic Language),
FORTRAN, and COBOL, with a liberal sprinkling of new facilities.
ALGOL's principal contribution was block structure and nested
constructs, while FORTRAN contributed scientific processing and
COBOL added commercial facilities. This combination produced a
large, unwieldy language with twists and nuances that can trap
the unwary programmer. Nevertheless, PL/1 was quite
comprehensive, and it served as the basis for uncounted numbers
of application programs on large systems. One noted use of PL/1
was in the implementation of the Multics operating system at
MIT, under Project MAC.
In 1976, an ANSI (American National Standards Institute)
committee produced a standard language definition for PL/1. The
standard is an implementation guide for compiler writers, and it
precisely defines the form and function of each PL/1 statement.
Aware that PL/1 was too large and complicated, the committee
produced a smaller version for minicomputers, called Subset G.
This new language excluded the redundancies and pitfalls of full
PL/1, but retained the useful application programming features.
Recently approved by ANSI, Subset G has given new life to PL/1,
with manufacturer support for the Data General Eclipse and
MV/8000 computers, Prime computers, Wang machines, and DEC's
popular VAX computer.
Strangely, the refinements found in application languages follow
those of hardware and operating systems. Large, cumbersome
languages have been rejected in favor of simple, Spartan
programming systems that are consistent in their design. The
resulting languages are easier to implement, simpler to
comprehend, and allow straightforward program composition.
System Languages
----------------
A system language is a high-level machine-dependent programming
language used to implement so-called "system software,"
including operating systems, text editors, debuggers,
interpreters, and compilers. In the early days of computing,
virtually all system software was implemented in assembly
language. One revolutionary machine, the Burroughs B5500, used a
variant of ALGOL-60 as its only system programming tool, and
appeared in the early 1960s. The machine was a commercial
success against the other major mainframes, proving that
assemblers were no longer necessary. Many successful system
languages followed Burroughs' ALGOL, including the C language,
produced at Bell Laboratories in the late 1960s, which served as
the basis for the UNIX operating system.
A system language, by definition, matches the architecture of a
particular machine or class of machines; all facilities of the
machine are accessible in the language, and the language
contains no nontrivial extensions beyond the basic machine
capabilities. The benefit is that a compiler for the system
language is easy to implement and transport from machine to
machine, as long as the architecture of each machine is similar.
Further, a system language requires little runtime support since
application facilities, such as extensive I/O (input/output)
processing, are not generally embodied in the language.
Refinements in system languages are made by increasing their
usability. Their acceptance as replacements for assembly
languages is encouraging. Today, one can publicly admit that
system software is implemented in a high-level language without
implying that it must be rewritten in assembly language to be
effective.
PL/M: The Base for CP/M
-----------------------
In 1972, MAA (Microcomputer Applications Associates), the
predecessor of Digital Research, consulted with the small,
aspiring microprocessor division of a semiconductor memory
company called Intel Corporation. MAA defined and implemented a
new systems-programming language, called PL/M (Programming
Language for Microcomputers), to replace assembly-language
programming for Intel's 8-bit microprocessor. PL/M is a
refinement of the XPL compiler-writing language which is, in
turn, a language with elements from Burroughs Corporation's
ALGOL and the full set of PL/1.
The first substantial program written by MAA using PL/M was a
paper-tape editor for the 8008 microprocessor, which later
became the CP/M program editor, called ED. PL/M is a commercial
success for Intel Corporation and, although licensing policies
have limited its general accessibility, it has become the
standard language of the Intel microprocessor world, with
implementations for the 8080, 8085, and 8086 families.
MAA also proposed a companion operating system, called CP/M
(Control Program for Microcomputers), which would form the basis
for resident PL/M programming. The need for CP/M was obvious;
8080-based computers with 16 K bytes of main memory could be
combined with Shugart's new (at that time) floppy disk drives to
serve as development systems. For the first time, it was
feasible to dedicate a reasonably powerful computer to the
support of a single engineer. But the use of PL/M on larger
timesharing computers was considered sufficient, and the CP/M
idea was rejected.
Operating Systems
-----------------
Operating systems, too, have become more refined. But why do we
have operating systems at all? In the 1960s, we used expensive
mainframes with power-hungry central processors and magnetic-
core memory. Downtime for complicated card readers, printers,
and backup data-storage devices was high, requiring constant
maintenance. A card-oriented "batch" operating system provided
two functions. First, it allocated processor time, memory, and
peripherals to application programs, in an attempt to utilize
each expensive component to its fullest. Second, common I/O
subroutines were a part of the operating system, to avoid
duplication in each application program. In the early 1960s,
batch operating systems began to incorporate online terminals
that allowed the programmer to interact with the program -- this
is where things became interesting. With an online terminal, a
program could write a prompt message, read the data entered by
the operator, and write a response almost instantly.
The crude terminal systems evolved into today's timesharing
computers, where program interaction is the primary function,
with batch processing in the background. General Electric and
Digital Equipment Corporation led the way with BASIC-based 235
and multilingual PDP-10 computers. Countless timesharing
operating systems followed, including IBM's interactive APL and
CP/CMS, along with UNIX from Bell Laboratories. These
timesharing systems were the forerunners of personal computers:
all assumed that the hardware was too expensive to dedicate, so
each terminal becames an emulation of a single computer.
The CP/M Family
---------------
CP/M was, however, completed by MAA in 1974. It included a
single user file system designed to eliminate data loss in all
but the most unlikely situations, and used recoverable directory
information to determine storage allocation rather than a
traditional linked-list organization. The simplicity and
reliability of the file system was an important key to the
success of CP/M: file access to relatively slow floppy disks was
immediate, and disks could be changed without losing files or
mixing data records. And because CP/M is a Spartan system,
today's increased storage-media transfer rates simply improve
overall response. The refinements found in CP/M are based on its
simplicity, reliability, and a proper match with limited-
resource computers.
By the mid-1970s, CP/M added a new philosophy to operating
system design. CP/M had been implemented on several computer
systems, each having a different hardware interface. To
accommodate these varying hardware environments, CP/M was
decomposed into two parts: the invariant disk operating system
written in PL/M, and a small variant portion written in assembly
language. This separation allowed computer suppliers and end
users to adapt their own physical I/O drivers to the standard
CP/M product.
Hard-disk technology added yet another factor. CP/M customers
required support for disk drives ranging from single 5-inch
floppy disks to high-capacity Winchester disk drives. In
response, CP/M was totally redesigned in 1979 to become table-
driven. All disk-dependent parameters were moved from the
invariant disk operating system to tables in the variant
portion, to be filled in by the system implementer.
CP/M is now a multifunction program whose exact operation is
defined externally through tables and I/O subroutines. The
widespread use of CP/M is directly attributed to this
generality: CP/M becomes a special purpose operating system when
it is field-programmed to match an operating environment.
Through the efforts of system implementers who provide this
field-programming, CP/M is used worldwide in close to 200,000
installations with over 3000 different hardware configurations.
MP/M
----
As single-user CP/M became widely accepted, Digital Research
began to develop a new operating system for real-time
processing. The design called for a real-time nucleus to support
cooperating sequential processes, including a CP/M-compatible
file manager with terminal-handling capabilities. This operating
system, called MP/M (Multiprogramming Monitor for
Microcomputers), is a further refinement of the process model
found in Intel's RMX and National's Starplex. As a side effect,
the combination of MP/M's real-time nucleus with the terminal
handler and the CP/M file system produces a traditional
timesharing system with multiprogramming and multiterminal
features.
Timesharing allows programs to execute in increments of
processor time in a "lock-step" fashion. In a timesharing
context, a printer program, often called a spooler, might have
the task of printing a series of disk files which result from
program output. The spooler starts with a disk-file name and, by
using increments of processor time allocated by the real-time
nucleus, writes each line from the file to the printer. Upon
completion, the spooler obtains another disk-file name and
repeats the process. You can, for example, send the name of a
disk file to the spooler and, while the file is being printed,
edit another file in preparation for compilation. The spooler
and editor share processor time to complete their respective
tasks. In general, many such processes share processor time and
system resources.
MP/M process communication is performed through queues (or
waiting lines) managed by the nucleus. The spooler, for example,
reads file names from an input queue posted by another process
(which reads spooler command lines from the console). When the
spooler is busy printing a file, additional file names may enter
the input queue in a first-in first-out order.
Process synchronization through queuing mechanisms is common
place, but MP/M treats queues in a unique manner, simplifying
their use and decreasing queue management overhead. Queues are
treated as files: they are named symbolically, so that a queue
can be added dynamically. Like files, queues have queue control
blocks that are created, opened, deleted, written, and read. In
fact, the set of queue operations closely matches the file
functions of CP/M, so that MP/M provides a familiar programming
environment.
The implementation of queues is transparent to an operator or
system programmer, but it is important to MP/M's effective
operation on limited-resource computers. Queues are implemented
through three different data structures, depending upon the
message length. So-called "counting semaphores" count the
occurrence of an event with message length zero, and are
implemented as 16-bit tallies. Single-byte messages are
processed using a circular buffer. Similarly, queues containing
addresses are processed using circular buffers. In all other
cases, MP/M uses a general linked list, which requires
additional space and processing time. It is this sensitivity to
the capabilities of limited-resource computers that makes MP/M
effective: while real-time operating systems often incur 25 to
40% overhead, MP/M has been streamlined to increase available
compute time by 7% over single-user CP/M.
Like CP/M, MP/M is separated into variant and invariant
portions. The file-system interface is identical to that of
CP/M, with the addition of user-defined functions to handle non-
CP/M operations (such as control of the real-time clock). Field-
reconfiguration of MP/M allows a variety of device protocols,
including CP/M-style busy-wait loops, polled devices, and
interrupt-driven peripherals. In fact, the variety of interface
possibilities makes the MP/M implementer a true system-software
designer, since a fine-tuned MP/M system may operate
considerably faster than its initial implementation.
What are the refinements found in MP/M? First, it is a state-of-
the-art operating system based on current process-
synchronization technology and microprocessor real-time system
design philosophies. Process communication is conceptually
simple, and requires minimal overhead. Finally, it is the only
operating system of its type that can be field-tailored to match
almost any computer configuration.
CP/NET
------
CP/NET, introduced in late 1980, leads a series of network-
oriented operating systems that distribute operating system
functions throughout a network of non-homogeneous processors.
CP/NET connects CP/M requesters to MP/M servers through the use
of an arbitrary network protocol. Similar to CP/M and MP/M,
CP/NET consists of the invariant portion, along with a set of
field-reconfigurable subroutines that define the interface to a
particular network. For purposes of CP/NET, this interface need
only provide point-to-point data-packet transmission. Since the
actual data transmission media are unimportant to CP/NET, any
one of the number of standard protocols can be used, from low-
speed RS-232C through high-speed Ethernet. Physical connections
are also arbitrary, allowing active hub-star, ring, and common-
bus architectures.
The invariant portions of CP/NET operate under a standard CP/M
system to direct various system calls over the network to an
MP/M server. The MP/M server, in turn, responds to network
requests by simulating the actions of CP/M. This simulation is
transparent to an application program: any program operating
under standard CP/M operates properly in the network
environment.
Suppose, for example, that you wish to store common business
letters in a central data base under MP/M, and access these
letters from a CP/M-based word processor. You begin by assigning
one local disk drive to the MP/M master, using the CP/NET
interface. You then direct your word processing system to read
the particular letter on the assigned drive, causing the data to
be obtained from the server, rather than from the local disk.
After local update using your word processor, you can print the
result on your local printer, or optionally assign your listing
device to the network, for printing at the MP/M server.
CP/NET is accompanied by three related network operating
systems: CP/NOS, MP/NET, and MP/NOS. CP/NOS is, in effect, a
diskless CP/M, which can be stored in read-only memory, and that
operates with a console, memory, and network interface. MP/NET,
on the other hand, is a complete MP/M system with an embedded
network interface that, like CP/NET, allows local devices to be
reassigned to the network. MP/NET configurations allow MP/M
systems as both requesters and servers with CP/M requesters.
Finally, MP/NOS contains the real-time portion of MP/M without
local disk facilities. Like CP/NOS, MP/NOS performs all disk
functions through the network.
The interface protocol is publicly defined, so that non-MP/M or
non-CP/M systems can participate in network interactions. A
server interface for the VAX 11/780, for example, is under
preparation, so that it can perform I/O functions for a large
number of MP/M and CP/M requesters.
The principal advantage of CP/NET is that all CP/M-compatible
software becomes immediately available for operation in the
network environment, solving the problem that builders of
network hardware face: the total absence of application
software. Although the promise is there, networking is in its
infancy, and CP/NET is truly a software package awaiting the
evolution of suitable hardware.
PL/1: The Application Language
------------------------------
In 1978, Digital Research investigated the final level of
software support: application languages. One such language was
to be supported throughout the operating system product line,
and the choice would have to be a multipurpose language.
Further, the language would have to be an international
standard, to promote the generation of software by independent
vendors. Standard Pascal seemed a logical choice, but was
rejected for several reasons. First, Pascal is an ALGOL
derivative with scientific orientation. Commercial facilities in
the standard language are absent: decimal arithmetic, file
processing, string operations, and error exception handling were
essential. Further, separate compilation and initialization of
tables were not in the language. There was a temptation to
extend Pascal in order to include these features, but these
extensions would have defeated the benefits of standardization.
PL/1 Subset G was the obvious choice. It satisfied scientific
and commercial needs and, because of subset restrictions, was
consistent and easy to use. The project was a bit daring,
however, because Subset G was unknown in the computer community.
PL/1 was viewed as a large IBM-oriented language with huge,
inefficient compilers that required tremendous runtime support.
The Digital Research implementation of Subset G was started in
mid-1978 and completed two years later. The compiler is a three-
pass system written in PL/M. The first two passes are machine-
independent, and produce symbol tables and intermediate language
suitable for any target machine. The third pass is largely
machine-dependent, and is dedicated to code optimization and
final machine-code production. The compiler is accompanied by a
linkage editor (compatible with the Microsoft format), a program
librarian, a set of runtime subroutines, and a relocating macro
assembler.
Thus, PL/1 completes the final level of the inverted pyramid of
support tools. The message should be clear to the application
programmer: it is not the system language or the operating
system which is important in the production of a final
application. Rather, it is the availability of a standard,
widely accepted application language that can provide program
longevity. Once expressed in PL/1 Subset G, the program can be
transported through the CP/M family of operating systems to a
variety of minicomputer systems. Digital Research has a long-
term commitment to PL/1 support for popular operating systems
and processors.
New Processor Architectures
---------------------------
We have spent little time discussing processor refinements. What
is happening to our software tools as we augment our 8-bit
machines with the more powerful 16-bit processors? Will 16-bit
processors replace 8-bit machines, or are they simply a
temporary phenomenon in the transition to 32-bit machines?
There are several considerations when answering these questions.
First, 8-bit machines are economical to produce, their software
systems are mature, and they satisfy the needs of a substantial
computer base. Therefore, we can safely assume that 8-bit
machines are here to stay. Newer 16-bit machines are marginally
faster, but they have substantially more address space. To use
this additional address space, the computer must contain more
memory, which increases the computer system cost.
As system costs increase, the margin between low-end
minicomputers and high-end microcomputers diminishes, placing
microcomputer hardware and software manufacturers, such as
ourselves, in direct competition with major minicomputer
manufacturers. The 16-bit machines, by their nature, introduce
memory segmentation problems that are not present in 32-bit
processors.
Finally, we should note that 16-bit minicomputers are already
outmoded, and all serious manufacturers are pushing 32-bit
machines. This leads to the following conclusion: if we are
tracking the minicomputer world, we can assume that the future
will be with the 32-bit processors.
Currently, however, 32-bit machines are not available in
quantity. Even when they are available, there will be delays
while manufacturers tool up for production. At the moment, the
16-bit processors offer an intermediate solution. Digital
Research has provided initial support for Intel's 16-bit
machines (iAPX-186 and iAPX-286) which are versions of its 8086
product line. Intel provided PL/M-86, rehosted from the 8080
line, which was used by Digital Research to generate CP/M-86 and
MP/M-86. In both cases, the fundamental design remains basically
the same as that of the 8-bit version, with the addition of
memory management and enhancements to the file system that match
new computing resources. A familiar program environment is
retained, so that program conversion is simplified.
CP/NET and related network software will be available sometime
this year. Intel's 8087 (an arithmetic coprocessor for the 8086)
is of particular interest, since it directly supports binary and
decimal operations, which substantially increase PL/M-86
execution speed.
In addition to the 8086, the CP/M family will be adapted to the
16-bit machines that prove popular, with special interest in the
32-bit architectures as they become available. During this
development and rehosting, however, the 8-bit processors will
continue to be supported with new tools and facilities, since
this constitutes, without doubt, our best customer base for some
time to come.
Software Vendors
----------------
We have concerned ourselves with three levels of software tools
that support the most important level: the application programs.
A major reason for CP/M's popularity is the general availability
of good application software. At last count, there were about
500 commercially available CP/M-compatible software products.
Through the combined efforts of CP/M distributors, independent
vendors, and CP/M users, we are participating in a software
commodity market, with quality and variety that is unequaled by
any minicomputer or mainframe manufacturer. The large CP/M
customer base allows a vendor to produce and support a software
package at low end-user cost. This increases the customer base,
drawing more vendors with lower cost good-quality products. This
cyclic effect is, today, solving the "software crunch."
The tools are available, and it is the responsibility of
independent software vendors to continue developing their own
specialized markets. In this way, computer software technology
will reach virtually all application areas where low-cost,
reliable computing is required. Refinements? My friend, they are
up to you.
EOF
elaich
$8fcf...@news.wanadoo.fr:
> MAA also proposed a companion operating system, called CP/M
> (Control Program for Microcomputers), which would form the basis
> for resident PL/M programming.
Straight from the horse's mouth. Control Program for Microcomputers, not
Control Program/Monitor.
--
"No sports writers were harmed during the making of this post. And what I
want to know is - why not?"
elaich
> Intel provided PL/M-86, rehosted from the 8080
> line, which was used by Digital Research to generate CP/M-86 and
> MP/M-86.
MP/M-86? I wasn't aware that it was ever released.
Cliff Bedore
>"French Luser" <Bill....@microsoft.com> wrote in news:428ae8cb$0$25054
>$8fcf...@news.wanadoo.fr:
>
>
>
>> Intel provided PL/M-86, rehosted from the 8080
>>line, which was used by Digital Research to generate CP/M-86 and
>>MP/M-86.
>>
>>
>
>MP/M-86? I wasn't aware that it was ever released.
>
>
>
Oh yeah. We used it before CCPM-86 was released. From the user's
perspective, we didn't see a big difference between the two (i.e.
Wordstar, dBaseII, etc all ran the same.) I'm sure some of the more
technically astute here can point out the differences.
Cliff
French Luser
Have a few minutes free, today.
I received a message from a total unknown
containing the text of the BYTE article of
Gary Kildall.
The only explanation given was that it contained
the meaning of CP/M.
Don't ask me why, but, using the "Edition"
button, I "Search"ed for CP/VMS... "End
of file reached" (meaning: string not found)
I couldn't believe my eyes! ?!?!?!?!?!
(I must explain that, after finding (in the oldest
PL/M manual) the origin of the name "CP/M",
I noticed it in the BYTE article written by Gary
Kildall in 1981. Incredibly, I had not spotted it
15 years ago, when I retyped the article myself
(Old Timers remember that parts of it where
published in "The Computer Journal" when
Gary died.) Since I have papers from Gary
spanning 14 years, I can also reveal that
CP/VMS is only mentioned twice...)
So, I saved the file on my floppy and, that night,
I compared the text with the original BYTE article.
I found several things:
1) Many hyphens were missing. (Maybe the text
was not retyped, but processed by an OCR
machine?)
2) Many paragraphs were not ending as in
the original BYTE article.
3) Finally, the "Operating System" box
(containing the mention of CP/CMS)
was not present! QED!
So, I retyped it, did a quick on-screen
proof-reading (adding a few "," to make
the text clearer in meaning) and voila!
4) The order of the text/boxes was
different from the logical order of the
text (the one you have seen). Several
boxes were at the end of the text,
making the text, as a result, jump
from one subject to another...
You have seen the (quick) result of
one night of editing.
If there is a volunteer, the only right way
of proof-reading a text is by printing it,
then checking it line-by-line, word-by-word,
char-by-char, with the original...
(Example: while re-reading "Inefficient C"
from the cybercafe, I noticed "who wote
with their wallets"... Of course, I should
have retyped "vote"! Sigh!)
Good Luck! (Me, I am too busy at the moment.)
(The more I re-read this BYTE article, the more
I find it interesting. Gary was really a good
teacher.)
Yours Sincerely,
"French Luser"
Randy McLaughlin
news:ZMKdnVEOvdl...@megapath.net...
I have never seen any MP/M-86 version 1 nor MP/M-86 version 3 but there were
versions 2.0 & 2.1.
CCP/M-86 is an extension that was both a marketing tool to sell and a
different focus of implimentation.
With CCP/M you have a virtual console.
Randy
www.s100-manuals.com
Barry Watzman
(CompuPro) and Altos also sold it extensively for their systems.
wild bill
>"French Luser" <Bill....@microsoft.com> wrote in news:428ae8cb$0$25054
>$8fcf...@news.wanadoo.fr:
>
>> Intel provided PL/M-86, rehosted from the 8080
>> line, which was used by Digital Research to generate CP/M-86 and
>> MP/M-86.
>
>MP/M-86? I wasn't aware that it was ever released.
IBM was intent on doing business with microsoft.
In order to hamstring DRI, they gave them a very
strange NDA so they could 'talk about' the soon
to be IBM PC. According to the terms of that NDA,
EVERYTHING they discussed became IBM property.
Once DRI signed the NDA, IBM was more or less free
to do whatever they wanted in order to put an operating
system on the PC, including (obviously) buying or
licensing one from DRI, from ms, or from virtually
anybody else who had something to run.
What DRI had, and wanted to sell, was MP/M-86, which
obviously offered way too many features for a 'toy'
computer. It allowed serious computing applications
on inexpensive (read: not very profitable) micros.
From descriptions of that infamous 'meeting day', it's
clear IBM not only wasn't interested, they have never
mentioned that they were shown and offered MP/M-86.
What we all know is that IBM offered CP/M-86 in the
IBM style library slip case, languages section, for
approximately EIGHT TIMES the cost of the microsoft
product.
And it's worth noting that release was NEVER updated.
That means, no version 1.01, nada. Either it was perfect
from the start, or something funny was going on.
Okay, so, given the above, I suggest Kildall may very
well have said something like 'to hell with IBM; if they
don't want something more powerful, we'll skip them
and sell to every body else. They're not the only
computer company out there''.
There were dozens of IBM-compatibles hitting the
market - some could run ibmPC-dos, some could not.
But DRI needed to come up with something SIMILAR
to ms-dos, and CP/M-86 and MP/M-86 were just too
dissimilar; meaning, couldn't run ms-dos programs
Right there, you got to see actual INNOVATION.
The result was Concurrent DOS, and the features and
speed were at least two years ahead of anything ms
had at the time; DRI scored big, and ms was forced
into well-documented illegal acts and restraint of trade
in order to catch, then eliminate DRI as a competitor.
Bill
Barry Watzman
It became that way (very quickly, after the "Gary went flying"
incident), but it didn't start out that way.
The NDA given to DRI was the same as the one given to Microsoft. As
NDA's go, it was very onerous, but MS was willing to do "whatever it
took", and DRI was not. This is my clear recollection from
conversations with both Bill and Gary in conversations I had with both
of them more-or-less at the time (well, after the PC was formally
released, which was about a year after these specific events. But they
told the same story, more or less).
To the best of my knowledge, DRI was not actively trying to push MP/M-86
at that time (not that might not have been mentioned as a future
evolutionary path, but I think that your description mischaracterizes
the situation -- at the time of these negotiations, MP/M-86 didn't yet
exist).
But the real bottom line is that Gary and DR had no conception of the
importance and significance of the events that were transpiring, they
saw IBM as not terribly more significant than Heathkit / Zenith (which I
represented) or CompuPro or any of the other relatively large and
well-funded OEMs. Gates perceived the situation much more correctly. I
don't think that Gary was actually hostile to IBM, but Gates was willing
to kiss and suck and even swallow whatever IBM wanted, while Gary was
basically more ambivalent than anything else.
I was fairly involved with Gary and MP/M-86, and in fact my departure
from Zenith in mid-1983 occured to a small degree because I wanted
Zenith to implement MP/M-86 on the Z-100, and when Zenith didn't, I did
it myself and offered it for 3+ years. Concurrent {whatever}came much
later than the time period of the IBM PC introduction, more like 1984,
perhaps even 1985.
rip...@azonic.co.nz
> EVERYTHING they discussed became IBM property.
> Once DRI signed the NDA, ...
That was actually the point of the IBM going to MS, DRI would _not_
sign the NDA.
> What DRI had, and wanted to sell, was MP/M-86,
The original release of CP/M-86 was January 1981 *. The alleged
'lateness' of CP/M-86 was only that of the IBM-PC version because DRI
had not signed the NDA and thus did not have the ability to write the
BIOS code.
DRI sold appropriate products and MP/M-86 was for multiuser machines,
not 'PC's.
What DRI did want to sell, alongside CP/M-86, was Concurrent-CP/M-86
which was initially a single-user multi-tasking and was demonstrated
before the IBM-PC was announced.
> From descriptions of that infamous 'meeting day', it's
> clear IBM not only wasn't interested, they have never
> mentioned that they were shown and offered MP/M-86.
So what makes you think that DRI "wanted to sell it" or that IBM were
"not interested" ? Was it because of its complete absense from
discussions ?
What DRI did want to sell, alongside CP/M-86, was Concurrent-CP/M-86
which was initially a single-user multi-tasking and was demonstrated
before the IBM-PC was announced.
> What we all know is that IBM offered CP/M-86 in the IBM style
> library slip case, languages section, for
> approximately EIGHT TIMES the cost of the microsoft product.
> And it's worth noting that release was NEVER updated.
> That means, no version 1.01, nada. Either it was perfect
> from the start, or something funny was going on.
What "was going on" is that DRI had demonstrated that PC-DOS had DRI
copyright code in it and so IBM settled. The settlement included money,
DRI being able to clone PC-DOS (which is why it was not sued over
DR-DOS) and IBM selling CP/M-86 alongside PC-DOS. But IBM shafted DRI
by charging too high a price and never upgrading. DRI had to retail its
own version of CP/M-86 at a competitive price.
> I suggest Kildall may very well have said something like 'to hell
with IBM;
Wrong way around. IBM said 'we are forced to sell this so let's shaft
DRI over it".
> There were dozens of IBM-compatibles hitting the
> market - some could run ibmPC-dos, some could not.
Wrong time-line. There were many 8086 based machines which could run
MS-DOS (or SCP-DOS) or CP/M-86 and some already existed before the IBM,
but 'compatibles' did not appear until a new BIOS had been written.
> But DRI needed to come up with something SIMILAR to ms-dos, ...
> The result was Concurrent DOS, and the features and
> speed were at least two years ahead of anything ms
> had at the time; DRI scored big, and ms was forced
> into well-documented illegal acts and restraint of trade
> in order to catch, then eliminate DRI as a competitor.
You are confused. You probably meant DR-DOS 5 and, later, 6.
Concurrent-DOS did run some DOS programs but was a multiuser
multitasking system derived from MP/M-86, running on systems designed
for multi-user operation, and had no impact on MS-DOS or IBM-PC (though
it could be used on IBMs).
* Osborne/McGraw-Hill CP/M-86 User's Guide 1985
rip...@azonic.co.nz
Version 1.1 October 1981 *
2.1 March 1983
*Osborne/McGraw-Hill.
rip...@azonic.co.nz
> Oh yeah. We used it before CCPM-86 was released. From the user's
> perspective, we didn't see a big difference between the two ..
An MP/M-86 user could only start a new task by Ctrl-D to get a new
command prompt which effectively froze the background program, you had
to exit the new task to get back to the older one. Concurrent-CP/M-86
on a suitable terminal could switch between several running tasks.
rip...@azonic.co.nz
> later than the time period of the IBM PC introduction, more like
1984,
> perhaps even 1985.
The origianal demonstration of CCP/M-86 was around August 1981.
Releases:
1.0 (single user) September 1982
1.0 for XT mid 1983
3.1 (multiuser) February 1984
3.2 (PCDOS option) September 1984
Barry Watzman
IBM to Digital Research (Gary) to get the OS. But DRI's actions -- both
Gary's absence and their balking at the NDA -- so soured IBM on DR that
the relationship was totally destroyed in one meeting.
Concurrent didn't exist in January of 1981. Or January of 1982. Or
January of 1983. Or even, I think (but am not sure) January of 1984. I
think that some people here don't have their timelines on the products
right.
However, the OS being sold (whatever you think that it was or should
have been) was simply not the issue that killed the DRI-IBM
relationship. The issues were the way that DRI treated IBM and reacted
to IBM's demands. The "product", the software, whatever it was, was
secondary.
Barry Watzman
there could be many, limited only by memory). The other tasks
continued running in the background.
But MP/M-86 was intended to be used with multiple physical (RS-232)
terminals, while Concurrent was intended to have multiple virtual
consoles running in Windows on the main "PC" (although it also did
support RS-232 terminals).
Concurrent primarily "looks" somewhat like Windows, while MP/M-86
primarily looks like 1970's and 1980's "timesharing" on multiple
physical consoles.
Bill Leary
> An MP/M-86 user could only start a new task by Ctrl-D to get a new
> command prompt which effectively froze the background program, you had
> to exit the new task to get back to the older one.
That is not my recollection of using either MP/M-86 or -80.
I recall taking advantage of the multitasking abilities to compile in one
session while I went on editing the related documents in another.
One thing I recall about CCP/M vs. MP/M was that it was rather more convenient
using the former from a single terminal than the later, though I can't remember
now why that was.
- Bill
rip...@azonic.co.nz
What I said was that CP/M-86 was first released in January 1981. In
fact DRI adverts through 1981 carried CP/M-86, eg Byte August 1981
p247.
> Or January of 1982. Or
> January of 1983. Or even, I think (but am not sure) January of 1984.
I
> think that some people here don't have their timelines on the
products
> right.
Osborne McGraw-Hill has Concurrent-CP/M-86 for the IBM-PC released
September 1982, but it had been demonstrated the previous year.
Certainly DRI adverts in Byte during 1983 had CCP/M-86.
> However, the OS being sold (whatever you think that it was or should
> have been) was simply not the issue that killed the DRI-IBM
> relationship. The issues were the way that DRI treated IBM and
reacted
> to IBM's demands. The "product", the software, whatever it was, was
> secondary.
Exactly. DRI treated IBM as it treated every other company. The
"meeting'' was with the OEM Sales Manger, Gary's wife, as it would be
for anyone else. IBM did not deal with mere women. The lawyer advising
against signing the NDA was another slight.
'Winning' the case against PC-DOS was pure impudence and interfered
with IBM's new baby, so of course they should be despised.
rip...@azonic.co.nz
Perhaps you didn't realise that you could get a 2nd command prompt with
Ctrl-D.
>From DRI's MP/M II User's Guide (MP/M-86 is the same):
"""Ctrl-D - detaches the currently executing program from the console
at which the Ctrl-D is entered. If no program is executing, the Ctrl-D
reattches detached programs waiting for the console."""
A further section gives more details and advice.
A detached program could run until it tried to output to the console at
which point it would hang awaiting a reattach. This was used with
MicroFocus Fileshare program which would run as a background process to
give file sharing with record locks to CIS Cobol.
It was also slightly useful to check things in the system (such as
MPMSTAT) without having to go to a 2nd terminal.
> though I can't remember now why that was.
Multiple virtual screens and switching these.
I still have a descendant of CCP/M-86 (Real-32) running on my desk and
I use it every day.
rip...@azonic.co.nz
which
> there could be many, limited only by memory). The other tasks
> continued running in the background.
Only until they attempted to use the console as it had been detached.
Certainly a background task could run detached if it didn't attempt to
output anything to the console.
> But MP/M-86 was intended to be used with multiple physical (RS-232)
> terminals, while Concurrent was intended to have multiple virtual
> consoles running in Windows on the main "PC" (although it also did
> support RS-232 terminals).
My initial use of Concurrent-CP/M-86 was with ICL PC2 and Quattro which
only had serial terminals but they were ICL 6402 and 6404 (colour)
which gave 4 virtual consoles on each terminal. Early ones had to
refresh from the box on each switch but later ones were buffered and
switched instantly. (I still have several lying around here.)
Later I used ICL DRS300s which had terminals connected over a
'microlan' and these gave 4 virtual consoles each with optional
windowing (which no one used). (I have a stack of these too).
With Concurrent-DOS-386, Wyse 60 terminals fitted with 'PC-MODE' gave 4
virtual consoles with instant switching.
> Concurrent primarily "looks" somewhat like Windows, while MP/M-86
> primarily looks like 1970's and 1980's "timesharing" on multiple
> physical consoles.
That depends on the version and the implementation. Certainly MP/M
II/-86 looks just like a 25x80 'green screen', but so do most of the
implementations and even the latest derrivitive of CDOS - Real/32 on
both the main screen and the terminals because they dumped the windowed
console after CDOS 5.0.
OTOH Some versions (such as System Manager and Real-32) can run an
actual Windows 3.11 as just one of its tasks (or actually several - you
can run Windows 3.11 on a Wyse 370 over a serial link or a dialup).
Bill Leary
news:1116476119....@o13g2000cwo.googlegroups.com...
> > That is not my recollection of using either MP/M-86 or -80.
>
> Perhaps you didn't realise that you could get a 2nd command prompt with
> Ctrl-D.
>
> >From DRI's MP/M II User's Guide (MP/M-86 is the same):
>
> """Ctrl-D - detaches the currently executing program from the console
> at which the Ctrl-D is entered. If no program is executing, the Ctrl-D
> reattches detached programs waiting for the console."""
>
> A further section gives more details and advice.
>
> A detached program could run until it tried to output to the console at
> which point it would hang awaiting a reattach.
Yes, I think that was it. When I did the compiles I had to redirect the output
to a file so it wouldn't hang, then review it after it was done.
> This was used with
> MicroFocus Fileshare program which would run as a background process to
> give file sharing with record locks to CIS Cobol.
>
> It was also slightly useful to check things in the system (such as
> MPMSTAT) without having to go to a 2nd terminal.
>
> > though I can't remember now why that was.
>
> Multiple virtual screens and switching these.
Seems right. When I used CCP/M I didn't have to play any tricks to keep the
background job from stopping.
> I still have a descendant of CCP/M-86 (Real-32) running on my desk and
> I use it every day.
I've got a copy of CCP/M-86 around here somewhere. I used it quite a bit for a
while back then. I've thought, from time to time, of installing it and trying
it out on modern (fast) hardware to see how it feels today.
- Bill
Scott Moore
> IBM was intent on doing business with microsoft.
>
> In order to hamstring DRI, they gave them a very
> strange NDA so they could 'talk about' the soon
> to be IBM PC. According to the terms of that NDA,
> EVERYTHING they discussed became IBM property.
>
Sure, IBM wanted to avoid doing business with the only
company in existence with wide experience making operating
systems, in preference for a company that not only had
no experience doing that, but not a single line of code
in that direction, either.
IBM was a big bad corporation that wore suits. Bill Gates
saw farther into the future than any one of us, and he
won. Whatever else you don't like Gates for, his performance
of going from absolute zero in the operating systems world
to its undisputed king in a decade flat was nothing short
of an amazing business feat.
Scott Moore
>>According to the terms of that NDA,
>>EVERYTHING they discussed became IBM property.
>
>
>>Once DRI signed the NDA, ...
>
>
> That was actually the point of the IBM going to MS, DRI would _not_
> sign the NDA.
Yea, it was like IBM thought they were going to introduce a unique
product that would change the world.
rip...@azonic.co.nz
> Bill Gates saw farther into the future than any one of us,
Actually Bill Gates is not noted for his foresight ahead of events,
only in retrospect. In the first edition of his "The Way Ahead" there
was not one mention of the Internet, yet today many probably think he
invented it (a later edition added mention).
Edition 1 of Windows 95 had no Internet connection, yet OS/2 had it as
standard at the time.
Bill Gates saw GEM being demonstrated and then announced that MS will
'any day now' have its own GUI.
When there is a trend, Gates will run to the head of the crowd and yell
'follow me'. He is an oppotunist with enough money to buy in leading
products to make it appear that MS is innovative. Yeah, like with
'Bob'.
> Whatever else you don't like Gates for, his performance
> of going from absolute zero in the operating systems world
> to its undisputed king in a decade flat was nothing short
> of an amazing business feat.
Using a similar set of ethics Adolf went from a beer hall to master of
90% of europe in the same time. I presume that you you admire him for
that too.
Kelly Hall
> Edition 1 of Windows 95 had no Internet connection, yet OS/2 had it as
> standard at the time.
I ran the trumpet TCP/IP stack on 3.11 so I'm pretty sure I had a socket
interface in win95. I think it's time you layed off the pipeweed,
Frodo; it's starting to affect your memory.
> When there is a trend, Gates will run to the head of the crowd and yell
> 'follow me'. He is an oppotunist with enough money to buy in leading
> products to make it appear that MS is innovative. Yeah, like with
> 'Bob'.
It's a careful strategy to entice smaller companies to take the risks
inherent in bringing new technologies to the marketplace. Let them
figure out what works and what doesn't then buy/steal from the survivors.
> Using a similar set of ethics Adolf went from a beer hall to master of
> 90% of europe in the same time. I presume that you you admire him for
> that too.
*splash* This thread has officially jumped the shark.
Kelly
rip...@azonic.co.nz
as
>> standard at the time.
> I ran the trumpet TCP/IP stack on 3.11 so I'm pretty sure I had a
socket
interface in win95.
You may want to note that 'Trumpet' is not the same company as
Microsoft and did not put its product on Windows issue disks.
Gates did not notice the future was the Internet and when he eventually
did he tried to replace it with MSN (the original one which was not
connected to the internet and was for Win95 users only) which was a
failure.
The claim was "Bill Gates saw farther into the future than any one of
us,", yet there are dozens of examples where he was scrambling to keep
up. Just a couple of weeks ago he was announcing that Longhorn will
eventually have features that are already in Tiger and KDE and making
it sound like he was being innovative. Only Windows users were fooled.
Kelly Hall
> You may want to note that 'Trumpet' is not the same company as
> Microsoft and did not put its product on Windows issue disks.
Trumpet was just the (free) stack that I used because I didn't have the
Microsoft stack.
> Gates did not notice the future was the Internet and when he eventually
> did he tried to replace it with MSN (the original one which was not
> connected to the internet and was for Win95 users only) which was a
> failure.
You're confusing the consumer-level product (MSN) with the underlying
developer-level API.
I don't listen to pie-in-the-sky hot air from computer company execs;
and I don't care to refute your opinions. I chimed in because you got
the facts wrong about 'internet' support in legacy Windows products.
Kelly
Scott Moore
>
>
>>Bill Gates saw farther into the future than any one of us,
>
>
> Actually Bill Gates is not noted for his foresight ahead of events,
> only in retrospect. In the first edition of his "The Way Ahead" there
> was not one mention of the Internet, yet today many probably think he
> invented it (a later edition added mention).
>
> Edition 1 of Windows 95 had no Internet connection, yet OS/2 had it as
> standard at the time.
>
> Bill Gates saw GEM being demonstrated and then announced that MS will
> 'any day now' have its own GUI.
>
> When there is a trend, Gates will run to the head of the crowd and yell
> 'follow me'. He is an oppotunist with enough money to buy in leading
> products to make it appear that MS is innovative. Yeah, like with
> 'Bob'.
>
Gates saw several things that Kindall did not:
1. That IBM would be successful in the microcomputer market, and bet on
them accordingly. Don't laugh. Back then many people were predicting that
IBM would fail.
2. That the operating system for the new machine would be a big, big deal.
>
>>Whatever else you don't like Gates for, his performance
>>of going from absolute zero in the operating systems world
>>to its undisputed king in a decade flat was nothing short
>>of an amazing business feat.
>
>
> Using a similar set of ethics Adolf went from a beer hall to master of
> 90% of europe in the same time. I presume that you you admire him for
> that too.
>
I don't equate gates with Hitler. You should be ashamed to say that of
anyone. Hitler was a mass murderer. Calling anyone that who does not
deserve it is offensive. I studied history, I know more about Hitler
(and Stallin) than you ever will.
I think you shame the memory of Hitlers victims, as well.
rip...@azonic.co.nz
> the facts wrong about 'internet' support in legacy Windows products.
Excuse me, but my claim was:
"""Edition 1 of Windows 95 had no Internet connection"""
Your 'refuting" was that '_other_companies_ could do internet connect'.
> Trumpet was just the (free) stack that I used because I didn't have
the
> Microsoft stack.
Exactly. That was my point. You used Trumpet (or IBM or a couple of
others) because Microsoft didn't supply a mechanism to connect to the
internet until later. It was in Win95 OSR2, but was not in the first
release.
The later MS Plus! pack added internet access and IE to the original
Win95.
After Win95 was released MS realised that MSN was not going to replace
the Internet and so put out the Plus! pack, added this to OSR2 and
built IE and TCP/IP for Win3.11. They reinvented MSN as an ISP and
hoped everyone would forget they didn't see the internet coming.
They obviously fooled you, perhaps you only had Win95 OSR2.
rip...@azonic.co.nz
on
> them accordingly. Don't laugh. Back then many people were predicting
that
> IBM would fail.
IBM originally planned on making 50,000 computers. Their target market
was their own sites where Apple IIs running Visicalc were appearing.
IBM wanted a machine that would match the Apple II and run the same
software (BASIC, Visicalc) and could also act as a terminal and should
be thrown together using existing designs and parts, such as the System
23 Motherboard and obsolete (even at that time) 8052 serial ports.
Gates was opportunistic in that he knew SCP had a CP/M knockoff and he
took out a licence for it (he _later_ bought it outright). The licence
allowed him to sell it to any number of OEM customers and IBM was just
one of those. Another was Victor with the Sirius that predated the IBM
release.
You may have an argument about 'how gates predicted the future' if
there wasn't clear evidence that he wasn't putting all his eggs in the
IBM basket and was selling to anyone and everyone. Basically, Gates
didn't care whether IBM, or DEC, or Apple, or Victor or homebuilt was
successful, as long as they bought his software.
It is the same today, he doesn't care whether it is Intel or AMD, Dell
or Gateway as long as they send him money. He is even betting on IBM
Cell now. He is opportunistic not visionary.
> 2. That the operating system for the new machine would be a big, big
deal.
MS already had Xenix as an OS that it was using internally and was
selling to multiuser sites. The MS plans were for DOS to become a low
end to the MS 'family' of OSes. DOS 2 features were based on Xenix
features, and for example, the device drivers in 2 were loosely based
on Xenix drivers. It is clear that Gates initially treated DOS as a
low-end entry level for the 'family' of systems.
For example MultiPlan (the forerunner of Excel) was avilable on Xenix
and DOS and this was being promoted well after DOS 2 was released.
It wasn't DOS that made IBM the machine of choice, MS-DOS was available
of almost every 8086/8088 machine being made, it was Lotus. MS-DOS did
not beome IBM specific until 5, some 10 years after the original
release.
The argument that 'Gates saw that IBM would be successful an gambled on
it' is nonsense. It was not until 10 years later that MS became IBM-PC
specific, and it also made MIPS, Alpha and other non-IBM stuff. Even
Windows 1 would run on non-IBM machines if the manufacturer built the
low level video drivers.
> Gates saw several things that Kindall did not:
Kildall had developed the portable micro OS market and had developed a
multiuser OS since 1978 to give a 'family' of compatible operating
systems across 8bit and 16 bit machines.
Gates had been trying to catch up by buying in a Unix licence to match
up to MP/M and by buying in QDOS from SCP to match CP/M (which it did
almost exactly, too exactly). He then tried to make them into a family
by adding Xenix like features to DOS (and vv).
Gates is almost always late to innovation. He announced Windows when
there was not yet a project when he saw GEM demonstrated. He bought in
Lattice C when customers were sucking up C and not buying MS Pascal. He
bought in Mosaic and had Spyglass make IE when Netscape was what people
wanted.
It took Gates nearly 2 years to catch up with DR-DOS 5 and nearly a
year to catch up to 6. IBM had written PC-DOS 4 with >32MByte disks
because MS had not kept up when Compaq and Wyse had modified their own
OEM versions, and DRI OSes had no such limit for years. MS LANManager
was bought in from IBM (for DOS and Unix/Xenix) to catch up to Novell
(it failed to).
Things which DRI had and MS was in catch up mode:
Single user OS (CP/M CP/M-86)
Multi user OS (MP/M family)
Networking (CP/Net)
Graphics UI (GEM)
C language
Hard drive support (MS-DOS 1 lacked this)
>32 MByte drives (IBM had to write the code)
CDs *
DOS 5 (took nearly 2 years)
DOS 6 (took nearly a year)
Things which were MS innovations:
MS Pascal
MSX
MSX-DOS (actually it was just a CP/M clone)
Bob
Per box pricing
Bundling (MS-DOS+Windows at Windows price)
Ripping off Spyglass because IE was never _sold_.
'commitment' discounts (ie 100% MS)
Anti-trust violations
Gates is only a visionary in retrospect, and after a rewrite of
history.
* """Some good ideas came out of the skunkworks, although the best
mostly came from Gary. He did groundbreaking work on CD-ROM software
and on interfacing computers and video disks. A company, KnowledgeSet,
came out of that work. So did a CD-ROM-based Grolier's Encyclopedia, a
product that showed everyone how to do CD-ROM content. Microsoft's
later enviable position in the CD-ROM content market owes a lot to Gary
Kildall's good ideas and Bill Gates' ability to spot a good idea and
pounce on it."""
Axel Berger
>I ran the trumpet TCP/IP stack on 3.11
Me too, but it definitely was NOT part of the OS but third party
software that often came bundled with proprietary browsers.
At the time someone extracted the stack out of the IE package to let me
run Netscape with it but it was not smooth at the time (you had to do
things in just the right order) and I do not know how to recreate it
for my W3.11 notebook (486/66, 8MB).
Anonymous Guy
On 2005-05-20 Scott Moore said:
> I studied history, I know more about Hitler (and Stallin)
> than you ever will.
Thatt's fascinnating, Sccott. Pleasse telll uss morre aboout
youur hisstory backkground.
Frank McConnell
riplin wrote:
> Edition 1 of Windows 95 had no Internet connection
My recollection differs. So does my perspective. In 1994 and 1995, I
was working for The Wollongong Group doing a TCP/IP stack for Windows
3.1/3.11 enhanced mode. You can blame me for the PathWay Runtime 4.0
TWGTCP.386 and WINSOCK.DLL, I was mostly responsible for those bits.
The Winsock 1.1 spec was current, and there were several TCP/IP stack
vendors for DOS/Windows 3.x who supplied WINSOCK.DLL for their stacks.
They all had subtle differences, both from each other and from the
Berkeley sockets API from which they were derived. I did do some work
toward making the PathWay WINSOCK.DLL work better with applications
developed for other Winsock implementations or ported from Berkeley
sockets.
Sometime in mid 1994 the PathWay Runtime team (myself included) had a
meeting with a Microsoft representative, initials DB, who looked at us
a bit funny and asked whether we understood that Chicago was going to
ship with a TCP/IP stack. Yes, we knew that, and that what we were
working on had a short lifetime in the market (although we were still
supporting it in early 1999, to my surprise!).
And by 1995 there was an effort under way, facilitated by Intel with
some assistance and effort from FTP Software, Novell, Cisco, Digital
Equipment Corporation, Microsoft, and others to clean up the Winsock
spec, better document some of the unwritten assumptions, expand the
API with additional functionality, and upgrade the architecture to
allow non-TCP/IP transport providers to be plugged in underneath
Winsock. This work resulted in a cleaned-up revision of the Winsock
1.1 document (though it seems to have gone down the memory hole since
then) and the slightly later Winsock 2.0 API and service (transport?)
provider interface specifications.
The first release of Windows 95 included a TCP/IP stack, NDIS driver
interface, SLIP and PPP support, and Winsock 1.1-I-think API
implementation. There was a later release of a Winsock 2.0 that could
be installed atop this on Windows 95, and I believe later releases of
Windows 95 included it. Memory fades though and I cannot tell you
exactly which release of Win95 first included a Winsock 2.0
WINSOCK.DLL.
It was possible to disable Microsoft's TCP/IP stack in that first
release of Win95 and install a third-party stack. One of my
co-workers at TWG did it. But Microsoft's stack was mostly good
enough and most users didn't see any need to replace it.
And of course, getting the Winsock API thoroughly documented made it
that much easier for Microsoft to do away with the need for TCP/IP
stack vendors for Windows, but that's OK, it let us get on with work
more beneficial to society, you know?
-Frank McConnell
rip...@azonic.co.nz
> interface, SLIP and PPP support,
Yes, I have the box here, but it connected to MSN only, a Win95 only
network run by MS. Win95 had EMail and so on, but only within MSN.
You could EMail other Win95 users and browse MSN. They soon gave up on
trying to restrict this and built a gateway for EMail.
The Win95 Plus! pack (I have a box here) came after Win95 and included
PPP suport to any ISP and IE allowing Internet access. OSR2 included
internet access.
This explains it all for those who can't remember:
elaich
@news1nwk.SFbay.Sun.COM:
> I don't equate gates with Hitler. You should be ashamed to say that of
> anyone. Hitler was a mass murderer. Calling anyone that who does not
> deserve it is offensive.
I don't approve of Gates' business practices, but he is one of the biggest
philanthropists in human history. At least he is giving a lot of it back.
Equating him with Hitler is mindboggling.
--
"No sports writers were harmed during the making of this post. And what I
want to know is - why not?"
glen herrmannsfeldt
> rip...@Azonic.co.nz wrote:
>
>> Edition 1 of Windows 95 had no Internet connection, yet OS/2 had it as
>> standard at the time.
> I ran the trumpet TCP/IP stack on 3.11 so I'm pretty sure I had a socket
> interface in win95. I think it's time you layed off the pipeweed,
> Frodo; it's starting to affect your memory.
Windows 95 never had a standard socket interface. That requires that
the socket descriptors and file descriptors have the same address space,
such that you can use read() or write() on them. It might be that NT
has that, but for 95 compatibility it isn't used.
-- glen
Anonymous Guy
On 2005-05-21 "Elaich" <a@b.c> said:
> I don't approve of Gates' business practices, but he is one
> of the biggest philanthropists in human history.
Actually, he isn't; neither on a total-dollar basis, nor on a
percentage-of-wealth basis.
And because his 'philanthropy' is doled out through a non-
profit foundation, the funding of that foundation is saving
him millions of dollars in federal taxes every year.
By the way, this philanthropy is a fairly recent phenomenon.
For nearly two decades, he kept ALL the money to himself.
It's only when he started running into major legal trouble
with the Guvvermint, and discovered that he had no 'friends'
in Congress, that his...uh, "contributing" begin in earnest.
But the more interesting story is in the fine print. If
you'll read closely, you'll discover that the bulk of his
non-political largesse usually takes the form of donations
of computer hardware and software; not cash.
The hardware, he gets at bulk wholesale prices. The software
(all M$ products, of course) costs him literally nothing.
The donation amounts, however -- the ones publicly reported --
reflect the -retail- value of the stuff he 'contributes.'
So overall, he's probably coming out way ahead on the whole
deal -- both financially, and public relations-wise.
Few things in this world are as they superficially appear to be.
That's one of the primary Facts o'Life, son.
> At least he is giving a lot of it back.
He's giving a tiny little bit of it back.
And, in doing so, he's also guaranteeing a continuing world-wide
market for M$ products -- as those unfortunate third-world victi...
errr, recipients of his largesse get hooked on the dumbed-down,
trailing-edge Mikro$loth brand of pointee-clickee mediocrity.
It's pretty much quid pro quo.
Hope this helps.
Guy Macon
Plllleeaasssse ddddoonnnn't..... Ppppoosssstttts aabbbboouutttt
Hhhhiittttlllleerrrr aannnndddd Ssssttttaalllllllllliinnnnn
aarrrrre ooffffffffff-tttttoopppppiiccccc hhhhheerrrrree.....
Barry Watzman
bought Windows 95, which was it's "release date", as part of a big
bundle promotion.
Cliff Bedore
>Scott Moore <samiams...@Sun.COM> wrote in news:d6lf10$c39$1
>@news1nwk.SFbay.Sun.COM:
>
>
>
>>I don't equate gates with Hitler. You should be ashamed to say that of
>>anyone. Hitler was a mass murderer. Calling anyone that who does not
>>deserve it is offensive.
>>
>>
>
>I don't approve of Gates' business practices, but he is one of the biggest philanthropists in human history. At least he is giving a lot of it back. Equating him with Hitler is mindboggling.
>
>
>
Let me see if I get this right. He runs competition out of business,
overcharges us for the software and when the government gets tough on
him, he takes the excess money he charged us and gives a little away and
he gets to be the good guy. What a world.
I'd much rather have the money back and give it to a charity of my
choice and get the tax break myself.
Cliff
CBFalconer
> elaich wrote:
>
... snip ...
>>
>> I don't approve of Gates' business practices, but he is one of the
>> biggest philanthropists in human history. At least he is giving a
>> lot of it back. Equating him with Hitler is mindboggling.
>
> Let me see if I get this right. He runs competition out of
> business, overcharges us for the software and when the government
> gets tough on him, he takes the excess money he charged us and
> gives a little away and he gets to be the good guy. What a world.
>
> I'd much rather have the money back and give it to a charity of my
> choice and get the tax break myself.
Careful. You are on the dangerous edge of being unpatriotic here.
Taxes should be imposed on the rank and file, and not annoy the
wealthy. Similarly medical and retirement benefits should be
shrunk, oil should be spilled on the wild places, and roads should
be cut everywhere. This is policy from our glorious leader and
cohorts. You could be dumped into the Gitmo Gulag and held
indefinitely as an enemy if you fail to agree. Rights - bah.
--
Some informative links:
news:news.announce.newusers
http://www.geocities.com/nnqweb/
http://www.catb.org/~esr/faqs/smart-questions.html
http://www.caliburn.nl/topposting.html
http://www.netmeister.org/news/learn2quote.html
rip...@azonic.co.nz
> he gets to be the good guy.
It is not quite that simple either. He has stock in the drug companies
that he is giving away the products of. They get a tax write off.
Meanwhile the countries are obliged not to let in cheaper 'pirate'
unlicenced copies of the drugs.
rip...@azonic.co.nz
At least some of it is recycled in-house machines which would otherwise
be landfill. As MS keeps up with the latest hardware the discards are
not very old - probably more recent than anything that I use.
Guy Macon
rip...@Azonic.co.nz wrote:
>At least some of it is recycled in-house machines which would otherwise
>be landfill. As MS keeps up with the latest hardware the discards are
>not very old - probably more recent than anything that I use.
I am shocked - shocked, I tell you - to find out that someone
who posts to comp.os.cpm uses older hardware! Who knew? :)
My theory is that the electrons get dizzy if they go faster than
a few MHz...
Bill Leary
>At least some of it is recycled in-house machines which would otherwise
>be landfill. As MS keeps up with the latest hardware the discards are
>not very old - probably more recent than anything that I use.
Along that line, MS has apparently found that they're not as able to push people
to new hardware as they once were. They're in the process of creating an
XP-line OS that will run on older machines.
http://news.yahoo.com/s/ap/20050518/ap_on_hi_te/techbits_windows_security/nc:1209
- Bill
rip...@azonic.co.nz
Note that it will only turn an old Win98 machine into a Windows
terminal to a Windows TSE server.
This is not so much to save money for people not wanting to buy all new
hardware but to ensure that all that they will spend goes to Microsoft
(except they will have to buy a server machine).
Axel Berger
>You should be ashamed to say that of anyone. Hitler was a mass murderer.
True, but beside the point. The point is that a semi educated failed
landscape painter could become the elected (!) prime minister of a
cultured democracy in rather a short time. Everyone only speaks about
what came after, but it is this period we need to keep in mind if we
want to prevent similar things in future.
And if it works in industry we need to try and understand how and why,
the day might come when it will work again in politics.
Randy McLaughlin
news:428F6C76...@yahoo.com...
I have the exact opposite political feelings of you and I am offended.
If you want to post so far off topic make it something we can all agree on
like religion rather than rant about things you don't know anything about.
Randy
CBFalconer
Well, I just broke down and ordered a used 450 Mhz P3 machine with
a 19" monitor and 128 Meg, which cost under $200 including about
$70 shipping. This may outperform my 80 Mhz 486 with 64 Meg. The
19 inch monitor was the deciding factor. Hope it works. I
estimate a speedup of about 15 to 20x.
wild bill
<Watzma...@neo.rr.com> wrote:
>Re: "IBM was intent on doing business with microsoft."
>The NDA given to DRI was the same as the one given to Microsoft. As
>NDA's go, it was very onerous, but MS was willing to do "whatever it
>took", and DRI was not.
DRI at that time HAD about 95% of the operating system business
for microchip computers. Not they were GOING TO HAVE, one day.
They already HAD it.
Gates and ms had one (of many) available versions of Basic and
a few other derived languages, nothing original, except maybe to
the extent that they ran on microcomputers.
DRI had something to loose (or, something to steal).
Gates and ms did NOT.
Nothing was ''the same''.
I do not think IBM EVER had any intention of doing business with DRI.
And I still have not had any answer to the question: Why is there
no mention of DRI anywhere in the IBM version of DR Logo?
Did IBM KNOW, or, as with John Draper, were they kept in the dark?
Bill
wild bill
<samiams...@Sun.COM> wrote:
>Gates saw several things that Kindall did not:
>
>1. That IBM would be successful in the microcomputer market, and bet on
>them accordingly. Don't laugh. Back then many people were predicting that
>IBM would fail.
That's pure bullshit. Kildall offered to sell him DRI. He (Kildall)
may have wanted $25 millions or so, but would probably have
taken something in the $10-$15M range. If your 'hero' knew even
a fraction as much about the business as you seem to think, he
would have saved himself several billion dollars right there.
Kildall got (reportedly) $160 million from Novell a short time later.
Gates knows business like the Cosa Nostra knows business.
He broke the law, plain and simple. The US government has
too many ball-less bureaucrats and almost no defenders of
the Public Interest left, so he's gotten away with it.
>2. That the operating system for the new machine would be a big, big deal.
See my answer to #1, above. He had no such idea. He just wanted
to sell IBM whatever he could for as long as he could, then punish
them when they tried to break away. He cost IBM hundreds of
millions, if not billions, plus the personal computer market, by
essentially tricking them into OS-2. Written in ASSEMBLY
language. For the 286 (meaning, real mode). IBM got screwed.
Gates sort of accidentally 'innovated' a handy 386 feature
Intel had been building into their chips, at just the right time.
Windows 3 was ms' first to run any protected mode code.
Not for all of it's parts/tasks/subs, but at least for some.
Haven't taken apart DRI products so I don't know (yet)
how or when they figured it out. Got work left to do....
More to the point, in answer to your #2, if he had really seen
that, he wouldn't have planted a field for Linux to harvest.
Many contributors to Open Source are fed by hatred of
gates' and microsoft business practices. Ms is damned if
they try to 'play fair', and damned to keep creating enemies
if they don't. If Ms doesn't find a way to deal with Open
Source, they are doomed sooner or later to fail in the
software business.
Bill
Tom Lake
> to sell IBM whatever he could for as long as he could, then punish
> them when they tried to break away. He cost IBM hundreds of
> millions, if not billions, plus the personal computer market, by
> essentially tricking them into OS-2. Written in ASSEMBLY
> language. For the 286 (meaning, real mode). IBM got screwed.
Well, you got that backward. According to memos which I personally saw, it
was IBM that insisted
OS/2 run on the '286. Gates really wanted to break away from its
limitations and Gordon Letwin
agreed that a '386-only version was needed. Since IBM was funding the thing,
they called the shots.
Tom Lake
primo
wrote:
And in so doing will create even more machines that run worse than
they did before.
Running the same piece of software on my P3-500 with win98se and on a
friends P4-3ghz with win XP, my p3 actually finishes first. Its sad
that XP slows a comp down that much. I won't be giving up on 98 till
they fix the problems with XP.
Strangely enough this Celeron 466 machine running win98se does the
same tasks (as well as 1000s of others) as the new OS is supposed to
handle.
My twiin to this machine died this week and will have to be replaced,
but the first thing that the new machine will get is a hard drive
transplant from the dead machine so that runs 98se as well.
rip...@azonic.co.nz
Probably not. Mary Gates (Bill's mother) had served on the United Way
board with IBM chairman John R. Opel and through connection an
instruction was passed down to 'go see Microsoft'. DRI was not part of
that 'directive' from the very top.
Kelly Hall
> Running the same piece of software on my P3-500 with win98se and on a
> friends P4-3ghz with win XP, my p3 actually finishes first. Its sad
> that XP slows a comp down that much. I won't be giving up on 98 till
> they fix the problems with XP.
About three years back I replaced the Win98 on my P3/500 with XP - I
found no decrease in speed and a huge improvement in stability and
uptime. The upgrade scripts ran well and did a great job.
I continue to run XP as my desktop OS at work and home - it seems stable
enough for me. But then again, I run FreeBSD for my real work and the
5.x series has enough problems that well, let's just say that I long for
my old UltraSparc box. Heck, I long for the old IBM 4341 I used in college.
Kelly
Guy Macon
primo wrote:
>Running the same piece of software on my P3-500 with win98se and on a
>friends P4-3ghz with win XP, my p3 actually finishes first. Its sad
>that XP slows a comp down that much. I won't be giving up on 98 till
>they fix the problems with XP.
Something else to look into:
Axel Berger
>I won't be giving up on 98 till they fix the problems with XP.
Well, I only ever "upgrade" if the new offers something that definitely
can't be done on the old AND that I really need or want. Meaning I
stick to old stuff for a looong time and when I change the new is to
a version signifcantly bigger than 1.0 and the worst bugs are out.
Everyone around here runs XP on machines much less powerful than my
biggest one while I stick to 98SE. People neither think nor care is
what I consistently observe.
Stephen Bendzick
>
> > 1. That IBM would be successful in the microcomputer market, and bet
> on
> > them accordingly. Don't laugh. Back then many people were predicting
> that
> > IBM would fail.
>
> was their own sites where Apple IIs running Visicalc were appearing.
> IBM wanted a machine that would match the Apple II and run the same
> software (BASIC, Visicalc) and could also act as a terminal and should
> be thrown together using existing designs and parts, such as the System
> 23 Motherboard and obsolete (even at that time) 8052 serial ports.
>
> Gates was opportunistic in that he knew SCP had a CP/M knockoff and he
> took out a licence for it (he _later_ bought it outright). The licence
> allowed him to sell it to any number of OEM customers and IBM was just
> one of those. Another was Victor with the Sirius that predated the IBM
> release.
>
> You may have an argument about 'how gates predicted the future' if
> there wasn't clear evidence that he wasn't putting all his eggs in the
> IBM basket and was selling to anyone and everyone. Basically, Gates
> didn't care whether IBM, or DEC, or Apple, or Victor or homebuilt was
> successful, as long as they bought his software.
>
> It is the same today, he doesn't care whether it is Intel or AMD, Dell
> or Gateway as long as they send him money. He is even betting on IBM
> Cell now. He is opportunistic not visionary.
>
> > 2. That the operating system for the new machine would be a big, big
> deal.
>
> selling to multiuser sites. The MS plans were for DOS to become a low
> end to the MS 'family' of OSes. DOS 2 features were based on Xenix
> features, and for example, the device drivers in 2 were loosely based
> on Xenix drivers. It is clear that Gates initially treated DOS as a
> low-end entry level for the 'family' of systems.
I was recently reading the article in _The_MS-DOS_Encyclopedia_ on
writing installable device drivers (as I was thinking of writing one),
and the communication protocol between DOS and the drivers clearly was
designed for a multitasking OS. DOS calls a "Strategy" routine in the
driver purely for the purpose of passing the driver a control data
structure, then it calls the actual operating routine--the "Interrupt"
routine--of the driver immediately after. The data structure contains
several status bits that you would expect to be polled signals, like
"Done" and "Busy". The driver must set the "Done" bit before returning,
when it is finished processing the requested function. (The "Busy" bit
is actually not used to indicate when the driver is busy, but to
indicate various conditions under a few driver function calls.) In a
single-tasking system like DOS, there is no reason to make separate
calls to pass data to the driver and to call the driver's operating
routine, and the "Done" bit serves no useful purpose--DOS cannot check
it until the driver finishes its work and returns, so it is always 0
when leaving DOS and 1 the next time DOS has control. (I wouldn't be
surprised if single-tasking DOS ignores it altogether.)
However, in a multitasking system, the Interrupt routines of several
active device drivers could be run simultaneously, and the Strategy
routines would be called as needed by the OS to command the drivers
without committing them processing time. The Done bit and other status
bits in the control data structures would provide real-time driver
status information that the OS (whenever it received processor time)
could use to detect when each command's execution was complete and its
return data was valid (which could be before the driver Interrupt
routine actually returned). Because of these features, a DOS
installable device driver could actually be run in a multitasking
environment without modification, unless inside its Interrupt routine it
required total control of the machine. (I believe the Encyclopedia
article explicitly asserted that a device driver must accept hardware
IRQs interrupting it asynchronously, so a driver that holds total
control by leaving interrupts disabled would be violating MS-DOS rules.
However, other behaviors presumptive of total control, such as changing
data values in low memory or manipulating general hardware registers,
might cause failure in a multitasking environment.) One caveat is that
the multitasking system must not execute the driver's Interrupt routine
when it does not have an active command request, since according to the
MS-DOS documentation, programmers can safely assume that a driver's
Interrupt routine will not be called without a command first being sent
by a previous call to the Strategy routine.
(Of course, in another article in the Encyclopedia, I also found the
assertion that later versions of MS-DOS [after v3.3] would be
multitasking; as I recall, it gave the impression that this development
was imminent. I know I've seen mention of this MS vaporware
announcement, with no source cited, on USENET before. I can post the
quote from the Encyclopedia if there's interest.)
> For example MultiPlan (the forerunner of Excel) was avilable on Xenix
> and DOS and this was being promoted well after DOS 2 was released.
>
> It wasn't DOS that made IBM the machine of choice, MS-DOS was available
> of almost every 8086/8088 machine being made, it was Lotus. MS-DOS did
> not beome IBM specific until 5, some 10 years after the original
> release.
I agree that MS-DOS, through at least version 3.3, was not IBM-specific;
MS clearly marketed it as an OS for any 8088/8086 system. For any given
DOS version, the kernel was compatible with any computer that ran an
8086/8088 CPU and was invariant among platforms; it was the MS-DOS BIOS
that made DOS work with specific hardware, whether IBM or proprietary,
and MS made development kits available for vendors to write/build the
MS-DOS BIOS (containing the resident device drivers) for a specific
system of theirs and integrate it with the standard kernel into a
ready-to-ship MS-DOS distribution version. One only has to look at all
the vendor labeled versions of early MS-DOS to see that this vendor
customization was well supported and very common. Zenith and Compaq
versions are two very common examples.
> The argument that 'Gates saw that IBM would be successful an gambled on
> it' is nonsense. It was not until 10 years later that MS became IBM-PC
> specific, and it also made MIPS, Alpha and other non-IBM stuff. Even
> Windows 1 would run on non-IBM machines if the manufacturer built the
> low level video drivers.
I agree that on evidence Bill Gates was into anything and everything
that might be the next big thing. He started out with BASIC for the
Altair; a big deal at the time, according to the editors of Popular
Electronics (from whom he learned about it) and the homebrew computer
community. Bear in mind this was the same Altair that Steve Jobs came
in to eclipse with the Apple I. But oh yeah, Gates sold BASIC for the
Apple I and II line. IBM came along and Gates produced DOS for them.
Then in 1984, Apple brought out the Macintosh. Gates and Microsoft
worked with Apple on software for that. Some years later, MS Windows on
IBM-compatible hardware became viable competition for the Mac. IBM
subsided, clone makers became industry leaders (Compaq, HP, Dell,
Gateway [2000], etc.), and MS moved with them to the systems they
evolved from the IBM architecture. The web became big; Gates tried to
displace it with MSN and then own it with IE (largely succeeding in the
latter). Video game consoles become huge business; MS introduces the
XBox. Palm computers (PDAs) gain popularity, so MS sells Windows CE.
If cell phones became viable general computing platforms (with decent
user I/O thruput), I'm sure MS would be there with a major systems
product. That's not visionary; that's wisely diversified business.
It's good corporate management, but it's not bold risk-taking.
Actually, I'd say that Microsoft has gotten to where it is by taking the
safest route possible. If it appears Gates started out by taking big
risks, it's my opinion that that's because business in the early
computer industry was intrinsically risky, as it is in any new industry.
While I'm at it, and aside, the very idea of Gates seeing that IBM would
be successful /and/ gambling on it is self-contradictory. If you truly
see ahead, then you're not gambling. Conversely, if it's a gamble, then
you're not sure. So either Gates was gambling or he was sure, but not
both.
> Kildall had developed the portable micro OS market and had developed a
> multiuser OS since 1978 to give a 'family' of compatible operating
> systems across 8bit and 16 bit machines.
Though I was born too late to know much of Gary Kildall and Digital
Research first hand (I was born in 1977), everything [credible] that I
have read about him indicates to me that he was a very intelligent
computer scientist and software designer with substantial (but not
tremendous) visionary leadership. While some things I have read are
undoubtedly biased in his favor (e.g. the collective content of this
newsgroup), I feel I have read enough varied and opposing sources and
perspectives to make a reasonably accurate rough judgment of his
character, role, and contribution. Through this, I have developed the
impression that CP/M, and all of Gary's other software, is solid and
respectable in both design and implementation. Also, from the primary
sources I have read, mainly the Byte article posted as the start of this
thread, I have difficulty believing that he was a poor businessman. He
may not have been the best entrepreneur (which, frankly, is no strike
against him by me), but he seems to have been in touch with his market
and to have been developing desired new software products without
alienating his established customer base (in contrast to MicroPro /
WordStar Int'l).
His business ethics also seem generally aligned with mine (which are
based on orthodox Catholic principles, to give you a rough idea). For
all these reasons, I partly wish he were still with us. On the other
hand, I don't grudge him his ticket out of here . . .
Stephen
rip...@azonic.co.nz
> designed for a multitasking OS.
Yes, it was. The Driver model for MS-DOS 2 was based on that for MS's
other Operating System: MS Xenix. Xenix was a licenced version of Unix
edition 7 for the 8086. Later MS sold this to SCO. MS wanted to
promote a 'family' of operating systems like DRI had and so
incorporated Xenix features into MS-DOS such as directories,
redirection, loadable drivers, etc.
> (Of course, in another article in the Encyclopedia, I also found the
> assertion that later versions of MS-DOS [after v3.3] would be
> multitasking; as I recall, it gave the impression that this development
> was imminent.
MS developed MS-DOS 4.0 (from 3.1) and 4.1 (from 3.2) which were
multitasking OSes (and not to be confused with the much later 4.01).
These were called 'European DOS' and were used for a time by Siemans,
Wang, ICL and Datapac (I worked for ICL at the time). MS scrapped this
before it got too far. See 'PC Interrupts' by Brown&Kyle for 'European
MS-DOS'.
Guy Macon
rip...@Azonic.co.nz wrote:
>MS developed MS-DOS 4.0 (from 3.1) and 4.1 (from 3.2) which were
>multitasking OSes (and not to be confused with the much later 4.01).
>These were called 'European DOS' and were used for a time by Siemans,
>Wang, ICL and Datapac (I worked for ICL at the time). MS scrapped this
>before it got too far. See 'PC Interrupts' by Brown&Kyle for 'European
>MS-DOS'.
"Also at around the same time, Microsoft was approached by Siemens
in Germany and asked to develop a multi-tasking version of DOS.
This they did, and it was actually quite a success. It was named
European MS-DOS 4.0, but because it was licensed by Siemens, did
not get a widespread distribution and so many people never got
to use it."
Tom Lake
> multitasking OSes (and not to be confused with the much later 4.01).
Actually, MS-DOS 4.0 was the only DOS developed by IBM. MS
had turned DOS development over to IBM when the famous "divorce" happened
and IBM decided to write its own, compatible DOS. MS licensed 4.0 from IBM.
DOS 4.0 was a turkey. Since IBM showed its great skill at microcomputer
OSes
(sarcasm here), MS then got control of DOS back for 5.0 and above.
Tom Lake
rip...@azonic.co.nz
> multitasking OSes (and not to be confused with the much later 4.01).
> Actually, MS-DOS 4.0 was the only DOS developed by IBM.
No. You _are_ confusing it with MS-DOS 4.01. Actually IBM developed
_PC_-DOS 4.0 which when handed back to MS was released as MS-DOS 4.01.
These have nothing to do with MS-DOS 4.0 and 4.1 which were
multi-tasking.
> MS
> had turned DOS development over to IBM when the famous "divorce" happened
> and IBM decided to write its own, compatible DOS.
No. That is not true. IBM developed PC-DOS 4.0 from 3.3 to incorporate
>32Mb partitions that Wyse and Compaq had already added to their OEM versions or MS-DOS. IBM also added some other features such as some changes to serial port handling.
PC-DOS 4.0 was _not_ 'IBM's own', it was _not_ a new compatible
development by IBM. It was based on MS-DOS 3.3 source code and was just
a modification of that.
> MS licensed 4.0 from IBM.
No, that is not true. MS and IBM always had cross licencing such that
developments by either party were available to the other. IBM had done
substancial rewriting of PC-DOS 1.0 to create 1.1 (some say to fulfill
the settlement with DRI) which was passed back to MS to be MS-DOS 1.25.
IBM also did much work on creating PC-DOS 2.1 these changes going into
MS-DOS 2.11. PC-DOS 4.0 used MS-DOS source.
> DOS 4.0 was a turkey.
Are you talking about PC-DOS 4.0, MS-DOS 4.01 or the multitasking
MS-DOS 4.0 ?
> Since IBM showed its great skill at microcomputer OSes
> (sarcasm here), MS then got control of DOS back for 5.0 and above.
MS always had control and received source code from IBM. MS didn't
bother to do any development because there was no competition. Compaq
and Wyse had added >32Mb partitions to their OEM MS-DOS 3.x and IBM
needed to compete with larger drives without having to use several
drive letters.
MS-DOS 5 was solely to catch up to DR-DOS 5 which had started to take
some market share from MS. MS only improves products when it needs to.
eg IE 6 was not going to be updated until Firefox started eating
share. In fact MS said that there would not be a new IE until
Longhorn, now they need to get IE 7 out to plug the holes.
Another 'feature' of MS-DOS 5 is that it was released on the 10th
anniversary of the release of PC-DOS because there was a 10 year
moritorium on MS selling retail versions for the IBM-PC.
rip...@azonic.co.nz
It wasn't particularly good, I did use it. The main problem was that it
was still real mode and limited to 640Kb which is just not good enough
for multitasking DOS software. Existing DOS software took no account of
other tasks and typically expected to stamp over all available memory
and use up 100% CPU time to check if a keyboard has been hit.
They attempted to make a 286 version and get developers to create
compatible applications but the 286 was a broken chip and most DOS
software did pointer arithmetic which meant it wasn't going to run in
286 mode.
MS-DOS 4.0 and 4.1 could run Windows 2, but it wasn't particularly
useful.
When Windows 286 and 3.0 could do sort of multi-tasking and this wasn't
stuck with having to run broken DOS software MS dumped 4.x.
So I don't know why anyone would think of it as a 'success'.
Richard Bonner
> rip...@Azonic.co.nz wrote:
> http://www.sylpher.com/dosuser/doshist.htm
*** Hmm, that is somewhat contrary to the story with which I am
familiar. I am under the impression that multitasking DOS was developed
under license in Europe, not by Microsoft itself. Microsoft kept it out
of North America and elsewhere because it did not want competition with
its Windows product, and was eventually able to kill it altogether.
Richard Bonner
http://www.chebucto.ca/~ak621/DOS/
rip...@azonic.co.nz
> under license in Europe, not by Microsoft itself.
The manuals for it were Microsoft. It was also used by others, such as
ICL and DataPac in Australia.
> Microsoft kept it out of North America
Wang used it.
> and elsewhere
DataPac in Australia developed a networked system that relied on MS-DOS
4.x for a large government department. When MS-DOS 4.x was dumped they
had to switch to another OS and chose DRI's Concurrent-DOS. They
became a DRI VAR (Value Added Reseller) and developed CDOS into System
Manager and added OzTerm and OzStation (network terminal) and many
enhancements.
> because it did not want competition with
> its Windows product,
Windows 2 ran on MS-DOS 4.0 and 4.1.
> and was eventually able to kill it altogether.
What killed it was that it wasn't very good - there was only 1
foreground process and it could run 'background tasks'. Certainly
Windows 286 and 3.0 did a better job of [sort of] multitasking because
they didn't try to use DOS programs to do so.
In any case MS started on developing MS-DOS 5 (not to be confused with
the much later MS-DOS 5) which became OS/2.
Kelly Hall
> MS-DOS 5 was solely to catch up to DR-DOS 5 which had started to take
> some market share from MS. MS only improves products when it needs to.
> eg IE 6 was not going to be updated until Firefox started eating
> share. In fact MS said that there would not be a new IE until
> Longhorn, now they need to get IE 7 out to plug the holes.
You make it sound like some vast conspiracy - no company improves their
product unless they need to. It's ridiculous to spend NRE funds unless
there's a direct payoff in sales.
Kelly
rip...@azonic.co.nz
> product unless they need to. It's ridiculous to spend NRE funds unless
> there's a direct payoff in sales.
Exactly. When MS manages to strangle its competition it has no need to
improve anything beyond the mediocre and can push the prices up to
increase its revenue. This additional revenue can be fed into other
markets (such as game consoles - current deficit approx. 400 million)
in order to attempt to take a monopoly postion there. The plan is to
kill off Nintendo and take enough share from Sony to make their unit
costs rise enough to hurt. MS will feed in a few billion more with
XBox 2, will tie up the game makers with exclusives and then suck the
revenue back when they are the only game in town.
They've done that before and will do it again, and again, if they can.
Is it a 'conspiracy'. No, a conspiracy has to be between several
partners, MS is doing it alone and fails on definition.
Anonymous Guy
On 2005-05-30 rip...@Azonic.co.nz said:
> Another 'feature' of MS-DOS 5 is that it was released on the
> 10th anniversary of the release of PC-DOS because there was a
> 10 year moritorium on MS selling retail versions for the IBM-PC.
It should also be noted that many of the outboard utilities included
with the M$-DOS 5.0 package were written by Central Point Software.
Some fairly substantial chunks of that DOS release were out-sourced.
I've never cared enough to find out why. Just assumed that M$ had
reached the limit of its creative, technical and programming powess.
In any case, the out-sourcing was a good move. M$-DOS 5.0 is rock-
solid; the best DOS (along with 3.3) ever released under the Mikro-
$loth imprimatur, IMHO.
Have a spare, still-shrink-wrapped OEM distribution package --
picked up at a rummage sale for $2.00 -- sitting on my software
shelf right now.
D'ya s'pose the floppies are still readable? ;)
rip...@azonic.co.nz
> with the M$-DOS 5.0 package were written by Central Point Software.
>
> Some fairly substantial chunks of that DOS release were out-sourced.
Many of which MS got for 'free' because MS convinced CPS that many
users would upgrade from the 'free' minimal versions included to the
full facility versions that cost money.
Of course most users found the included versions were good enough so
CPS simply contributed to MS's check list to compete with DR-DOS 5.
Note that DR-DOS 6 also included other's products but DR paid them a
cut of the DR-DOS price. DR-DOS 6's PC-KWIK was Multisoft, SuperStor
and AddStor were Addstor Inc.
> I've never cared enough to find out why. Just assumed that M$ had
> reached the limit of its creative, technical and programming powess.
MS's 'prowess' was always 'taking advantage of others' and they were as
exceptional as always.
> In any case, the out-sourcing was a good move. M$-DOS 5.0 is rock-
> solid; the best DOS (along with 3.3) ever released under the Mikro-
> $loth imprimatur,
It may have been the best MS-DOS to date but it still wasn't as good as
DR-DOS 5 * though it was around 20 months after it. And then DR-DOS 6
trumped it within a few months and it took MS nearly another year to
[almost] catch up.
* for example DR-DOS 5 made 286 machines usable with HiDOS. It also had
a much better command line editor and history, password protection for
files and directories.
Anonymous Guy
On 2005-05-31 rip...@Azonic.co.nz said:
> MS's 'prowess' was always 'taking advantage of others' and
> they were as exceptional as always.
I don't disagree with you. Personally, I have no M$ products
whatsoever on any of my computers.
Indeed, this message in being composed and transmitted under
DeviceLogics DR-DOS 8.0.
My only point was that in order to produce a decent, solid
DOS, Mikro$loth had to go "out-of-house."
Which pretty much tells the story.
French Luser
> Indeed, this message in being composed and transmitted under
> DeviceLogics DR-DOS 8.0.
Since this is the comp.os.cpm Newsgroup, and that
some people out there are still using DR-DOS 7.03,
could you (if you knew 7.03) tell us what is new,
and if there is any reason to upgrade? (Where?)
Yours Sincerely,
"French Luser"
Anonymous Guy
On 2005-06-03 Emmanuel Roche said:
> "Anonymous Guy" wrote:
>
> > ...DR-DOS 8.0.
>
> Since this is the comp.os.cpm Newsgroup, and that
> some people out there are still using DR-DOS 7.03,
> could you (if you knew 7.03) tell us what is new,
> and if there is any reason to upgrade? (Where?)
>
> Yours Sincerely,
> "French Luser"
Native pre-emptive multitasking, FAT32-capable,
large hard disk support.
$US 25.00 for a 1-user license.