Nick Demos writes:
>Wirt, while you make some valid points about multiple operating
>systems on one platform, this solution requires alot of extra resources,
>particularly if you are going to run them simultaneously. The POSIX
>type solution may be a better idea. It allows (or should IDEALLY
>allow) the easy conversion of a program written in a standard way to
>multiple platforms. That is the goal of POSIX. In the long run, I think
>it will be a better solution than multi-OS machines.
Nick is undoubtedly right, but I have become increasingly wary of this
solution as I have gotten older. especially given the times that we liven in.
Time is now measured in "Java Time," where hyperbole outpaces reality by a
factor of 20 to 1, and disappointment and fickleness turn the crowd agains a
particular product or process long before the first truly business-useful
applications appear.
Coming up with a common set of standards for code-portability such as POSIX,
as Nick mentions, is always the most efficient way to implement a common set
of operational standards. The only reservation that I have is that this has
been tried many times before -- with minimal success: the ANSI standards for
FORTRAN and COBOL , the development of Pascal, the language designed to usher
in the era of complete portability in code, and such forced-upon-you
standards as ADA. These standardization attempts don't work. Or if they do,
they don't last long.
The second, more psychologically debilitating attribute of hosting a
standardization attempt on your particular platform, especially if you are a
minority platform, is that you are harmed by the perception of never being
"with it," of always being late to the party, of not having people "port"
applications to your platform fast enough. Thus, there inevitably develops a
shrill chorus of people who want OODB, DCE, RPC, CORBA, NFS, DNS, or whatever
three- or four-letter acronym has captured the imagination of the moment.
In this era of compressed time, where we move from hyperbole to
disappointment to abandonment in a matter of weeks or months, this shrill
drumbeat becomes an impossible task for the vendor to manage, especially if
its primary considerations have always been in the past care and caution and
good design. A multi-OS system has the capacity to minimize much of that --
and that attribute alone may be its overarching reason for existence. It
would allow the best attributes of the well-designed systems to be expanded
more slowly and more carefully, while presenting in direct comparison to a
much broader audience, on the same box, the quality of solution that can be
built, if enough time and thought are given to the process.
I got the impression from Jeff Kell's posts that he might believe that a
multi-operating system would be implemented as a series of emulating shells
on one base O/S. That isn't the way multi-OS machines are likely to be built.
Rather, it is much more likely that any one OS would be constructed much more
like a piece of hardware, where you simply plug in a new set of discs and the
boot-up procedure recognizes which operating systems are present and which
are not.
Whether or not this ever comes to pass with the HP3000, the multi-OS
experiment is being conducted now by Macintosh and IBM. IBM and Apple will
soon have a common hardware platform that will allow them to run MacOS,
Windows 3.x, Windows NT, AIX, and OS/400, all on the same hardware. I would
not be surprised to see an announcement in the next several years announcing
a MacOS/Windows NT/AIX multi-OS system; they're already well on their way to
such a solution. I've included a public relations news release from Apple
below. Please pay special attention to the second and third paragraphs. If
this research is not simply marketing hype, then these results have
significant implications for the long-term prosperity of the HP3000.
The HP3000 and the Macintosh are very much alike in their core attributes:
ease-of-use, reliability, robustness, and low cost of operation. That
similarity was no accident; the HP3000 was the machine against which the
Macintosh was modeled. The HP3000 was "a machine that never cracked a smile
in its life," according to David Levy, the author of the Macintosh biography,
"Insanely Great," but was nonetheless the philosophical model used throughout
the design of the Macintosh. And because of that care in engineering quality,
it is undoubtedly no coincidence that the HP3000 and the Mac also share the
common attribute that their customers are extremely loyal and passionate
about their platforms -- and generally know the quality of solutions that
they have.
The great problem that HP and Apple faces, however, is that truly satisfied
customers are rarely good customers. Satisfied customers don't buy much; it
is the dissatisfied customers that spend money like water. We still use our
MacPluses we bought in 1984 or 1985 everyday. The PC's that we bought in that
same era were long ago dumped in some landfill and replaced with newer,
"better," and more powerful machines -- several times over now.
Denys pointed out that in 1995 Mac sales dropped from 9.5% of all PCs shipped
to only 6.5%. But what was not mentioned was that 1995 was the year of
Windows 95, when everybody, including us, had to go out and buy new machines
to run W95. That caused an abberational surge in PC sales -- that may well
harm overall PC sales in the coming years. It certainly made Macintosh sales,
which were otherwise up, shrink in proportion.
Nonetheless, the sales of PCs generated by W95 were significantly below
expectations and Intel's stock fell rather dramatically because of that
disappointment. Andy Grove, Intel's CEO, now openly talks about his recent
investments this spring in the start-up companies working on VRML products.
Grove sees these companies as being the next truly great resource hogs --
capable of generating the necessary public pressure to move to the next
generation of Intel processors. Without Intel finding some way to insure the
growth in demand for its increasingly faster processors, the company has no
where to go, and may actually begin to recede in overall revenues in a
satiated market.
There is more than a certain amount of manipulation in the marketplace, and
that manipulation is necessary for the continued growth of the well-known
market leaders. But it does lead to what I humbly call Wirt's Law: "The more
demand there is to buy and install the latest technology, the less practical,
productive, and profitable work is actually getting done."
The following is from an Apple press release (April 22):
====================================
The new PC Compatible products leverage upon key differentiators of
the Macintosh platform, such as ease-of-use, hardware/software
integration, built-in multimedia capabilities and the exceptional
price/performance value of PowerPC technology. In addition, these
products provide a high level of integration between the Mac OS and PC
environments, enabling users to run Mac OS and MS-DOS or Windows
software concurrently, easily switching between these environments
without rebooting. Users can copy and paste data between the Mac OS and
Windows environments, and even access Windows files and folders from the
Mac OS, and vice versa. System resources such as the floppy drive, hard
disk, CD-ROM drive, display, keyboard and mouse are shared to make this
a cost-effective and space-saving solution for customers with limited
budgets or desktop space. Customers can also access Macintosh-compatible
printers and modems from the MS-DOS and Windows environments.
Based on Apple's customer research, the company's previous
cross-platform Power Mac, the Power Macintosh 6100/66 DOS Compatible,
attracted x86-based PC owners to the Mac platform at almost three times
the rate of other Power Macs. In fact, the majority of Power Macintosh
6100/66 DOS Compatible customers were not replacing another Macintosh;
instead, they were either acquiring their first personal computer,
replacing a x86-based PC, or adding to their installed base of PCs.
Power Macintosh 6100/66 DOS Compatible customers spent, on average, 75%
of their time working within the Mac OS environment, and purchased as
many native Power Mac applications as other Power Mac customers.
"Customer research shows that Apple's previous cross platform
products have built additional momentum for the Mac platform and Mac
developers," said Dave Daetz, Apple's Worldwide Cross Platform Product
Line Manager. "And with today's introduction, we're taking a big step
forward providing hardware-based PC Compatibility, featuring Pentium and
586 processors across Apple's entire Power Mac family and empowering
customers to run more software on a Power Macintosh than on any other
mainstream PC."
PC Compatible Systems
Aimed at professional users in business, education and home
environments, the Power Macintosh 7200/120 PC Compatible is based on the
next-generation Power Macintosh 7200/120 announced today, but comes
preconfigured with one of two PC Compatibility Cards. The 7-inch
586 100 MHz version is an affordable, entry-level choice that includes
128K of secondary cache and 8MB RAM, which can be upgraded to 64MB.
Industry benchmarks show the speed of the 586 100MHz processor is
comparable to that of a Pentium 75Mhz. For customers who want a higher
level of performance, a 12-inch Pentium 100Mhz version is another
option. It features 256K of secondary cache and 8MB of RAM, which can be
upgraded to 72MB. The Power Macintosh 7200/120 PC Compatible also
includes 8MB of RAM for the Macintosh operating system.
Both versions incorporate an ATI Mach64 video controller, from ATI
Technologies Inc. of Toronto, Canada, for accelerated SVGA compatibility
up to 24-bits per pixel. All current Apple displays are supported, in
addition to many third-party Macintosh-compatible VGA and SVGA displays.
Customers can configure their Power Macintosh 7200/120 PC Compatible
with either one or two displays. When configured with a single display,
users hit a hot key to toggle between the Mac OS and PC environments;
the background environment continues running. When configured with two
displays, Power Mac 7200/120 PC Compatible users can keep an eye on both
environments concurrently at the full size and resolution supported by
the display and video controller. Fans of MS-DOS and Windows games will
also enjoy the built-in PC Game Port and 16-bit Sound Blaster Pro support.
The new PC Compatible system is designed to provide enterprise
customers in business or education environments with a wide range of
available networking options. Users can easily connect to a variety of
networks, including Novell Network SPX/IPX, TCP/IP and NETBEUI
protocols in Windows and DOS environments via a built-in Ethernet
connector and ODI and NDIS 2.0 drivers. Due to robust multi-node
support, users of the new Power Mac 7200/120 PC Compatible, with
the appropriate network access, can maintain two unique Internet
connections simultaneously -- from the Mac OS and Windows.
==================================
Wirt Atmar
|
