Gavin writes:
 
=============================
 
Denys goes on and on about something or other and then says:
>Buying a Mac is just too much of a stretch.
 
Ok, so buy a PowerPC running MacOS from IBM.
 
From today's Mercury News:
 
>IBM, FIVE YEARS after first entering an alliance with Apple
>Computer to develop  new technology, has agreed to license
>Apple's Macintosh operating system software.
 
G.
 
===============================
 
Or Denys could buy any of the new PowerPC Macs that also have a Pentium
(Pentium, PentiumPro, or 586) processor in them so that they can run DOS,
Windows 3.x, Windows 95, or OS/2, simultaneously with the MacOS. Switching
between environments is said to be accomplished with a single keystroke. Cut
and paste operations go both ways and files can be shared between the two
environments and are automatically converted in format.
 
Moreover, InfoWorld ran an article several weeks ago on its cover page
announcing that Apple had licensed Windows NT (presumably for the purposes of
dual operating system processing).
 
While the current crop of machines requires two processors in them (a 68K or
PowerPC processor for the Mac and an Intel processor for Windows), Apple and
IBM have been working for some time to create a common hardware platform that
will allow all of the operating systems to run (in more or less native mode)
on a common hardware platform.
 
The dual processor machines available now from Macintosh do share common RAM,
disc, keyboard, and CRT (although it is also possible to drive two separate
SVGA monitors from the same video card, one for MacOS and the other for
Windows).
 
I have thought for some time now that mulitple operating systems on one box
is a good idea -- and if Apple pulls this off (and they seem committed to it)
in a commercially successful manner, then there is going to develop some
substantial user pressure to have the same thing occur in all mid-range
machines.
 
An enormous amount of infrastructure accompanies any desired product or
process that appears on another platform. To bring over a rather foreign
process onto your desired platform, there are only two design alternatives:
either (i) emulate the other environment on whatever box you wish to move it
to (this is generally the poorer alternative because of the great number of
compromises that must inevitably be made), or (ii) re-create the second
environment in its entirety -- but within a completely "firewalled"
partitition -- on the desired hardware and provide for some very easy
"reach-through" for the files of interest that exist in the alternate
environment [along with (ideally invisible) signals and triggers to
synchronize processes in the two environments].
 
Doing this with the HP3000/HP9000 would seem a much easier task than that
that faces Apple, given that HP begins with a 100% guaranteed common hardware
platform, at least for MPE and UNIX. If HP should port Windows NT to PA-RISC,
as Denys recommends, then there is no reason why the resulting machine can't
be a tripartite device.
 
However, HP may not feel a great deal of pressure at the moment to port
Windows NT to PA-RISC, given the joint development of (a series of?) new
processors with Intel. HP-UX will certainly be moved to the new processors
and a Windows NT/HP-UX dual operating system machine would seem relatively
likely, especially if commercial success in the Mac makes the idea rather
common. If MPE is also moved to the Intel processors, then a (near
native-mode) common hardware platform will be established for all three
operating systems -- and all three O/S's could easily be available in one
box.
 
I have never been a fan of POSIX on the HP3000. It is inherently a solution
that can ultimately and completely satisfy no one. POSIX is not UNIX to a
true UNIX afficianado, and can never be. The true UNIX afficianado will never
see POSIX as anything other than a weak emulation of the "power" of UNIX. And
POSIX represents nothing but an unneccessary complication and
complexification of the HP3000 to an MPE-only user.
 
Success in any marketplace, just as it is in any evolutionary ecology,
ultimately lies in differentiation, not emulation. The emulating organism is
never seen as robust or competitive as the ideal it emulates -- and thus
loses out in every long-term competition. If MPE is to succeed, its only
course is to become to the easiest-to-use, most reliable, most resilient,
most robust business computing/database engine available, bar none. It's
already almost there, in my opinion. Trying to slowly meld UNIX into MPE will
erode MPE's inherent advantages -- and ultimately satisfy no one. The only
design path that ever truly succeeds is differentiation, where that
differentiation provides obvious and easily explainable advantages.
 
The only great trick associated with multiple operating systems lies in
insuring that the other O/S's (Windows, Windows NT, UNIX, etc) are allowed to
fail (as they are wont to do) gracefully so that MPE operations would not be
affected.
 
If Mac continues down this development path, where file "reach-through" is
easy, obvious (and perhaps invisible), then the advantages will be
significant: not every process "has" to be ported to the Mac; if a new
process becomes available on the Windows-based PC, it's simultaneously -- and
instantly -- available on the Mac, Denys' ODBC being just one example. The
same would be true of a Windows NT/HP-UX/MPE box. MPE would never lack for
the latest Internet process, but perhaps just as importantly, IMAGE databases
would be available to UNIX and Windows processes, complete with its extremely
robust file system and transaction manager..
 
There is a second pronounced advantage of such a design tack, also. This
second advantage is purely psychological (or political), but that does not
mean that it is not important. Denys is like most PC users; he's never used a
Mac (and not bloody likely to). The same is almost certainly true for most
UNIX users; they've never used an HP3000. However, the reverse is not true.
Almost all Mac users have used (and own) at least one PC; similarly most
HP3000 users keep a very keen eye on UNIX. These users know the alternative
operating systems reasonably well, have generally rejected them, and that
detailed knowledge of the alternate choices underlies much of the ardor and
loyalty that Mac and HP3000 users feel for their platforms. But if it were
possible for an HP3000 user to do something as simple as add a partitioned
disc to their machine to run UNIX and UNIX-processes, at any time, as an
appendage to their general MPE operation, without emulation, and be
guaranteed that they can run the same version of HP-UX that the "big boys"
run, then an enormous amount of political cover is provided for the strong
defense of the continued operation and expansion of the HP3000.
 
If HP has not yet given any consideration to the advantages of such a
multi-operating system design, it is perhaps time that they at least form an
investigatory committee to do so.
 
Wirt Atmar