I'd like just one really good operating system rather than numerous mediocre
ones combined in a MISS-MORE-MOST mess.

Jim







> Date:          Wed, 26 Aug 1998 15:19:21 EDT
> Reply-to:      [log in to unmask]
> From:          Wirt Atmar <[log in to unmask]>
> Subject:       HP3000s & Macs (was: QCTerm question (Part II) answers [long])
> To:            [log in to unmask]

> I earlier wrote:
>
> > As I replied to several people who mentioned that concern, the only real
> hope
> > is for the Mac OS/Motorola chip is at some point in time to become fully
> > compatible with the Intel processors, so that the Windows 98/NT OS'es can
> > operate in a fully partitioned MOS (multiple operating system) environment
> > with the Mac OS. Apple has been promising/threatening to do this for some
> > time, but they always seems to back off at the last moment. But if they do
> --
> > and they do it correctly -- then our code, and everyone else's, would run
> > without modification.
> >
> > If Apple doesn't do this, my most heartfelt advice is, for the sake of
> future
> > growth, is that the time has come to abandon the Mac. Please understand that
> > doesn't mean throwing the Macs you own out. They can continue to run until
> > the ends of their respective lives.
>
> Ted writes in response:
>
> > I'm amused and a bit amazed to hear this out of one of the folks who has
> > kept the HP3000 alive.  If there is a comparable group in the PC world
> > to the 3000 folks, it has to be the Mac aficionados.  Those folks have
> > been going through much the same thing we have.  They love the
> > machine, but the company seems clueless about how to sell it.  I
> > used to dislike the Mac because I'm a programmer and find it very
> > frustrating to not be able to get at the guts.  I don't want to be protected
> > from myself :-).  But for those who don't feel the need to tweak and
> > fiddle, the Mac has a lot of plusses and for certain markets,
> > the visual artists and musicians in particular, it is pretty much the only
> > machine to buy.  Particularly if Apple can get their act together (which, by
> > the by, I don't define as being able to pretend to be a Windows box ;-), I
> > expect the Mac to keep a fairly significant market share for many years
> > to come.
>
> There is much more to say on this subject than just the few words that I'll
> write here, but there is clearly a tendency to compare the HP3000 and Mac
> communities. The most obvious and overt similarity between the two is the
> enthusiasm of the two user groups. But the most obvious dissimilarity is that
> the HP3000 doesn't have a 30,000-ton elephant bearing down out there, creating
> a duplicate, but not quite duplicate, HP3000-like system. Unfortunately, the
> Mac does. Windows 95/98/NT is more Mac-like than it is DOS/UNIX-like -- and
> that puts the Mac in a terrible position.
>
> Apple knows it. Bill Gates knows it. And that's precisely why Bill invests a
> couple hundred million in Apple. His investment, his "commitment" to the Mac,
> allows him to flood the Mac world with Microsoft products that only make the
> decline and fall of the Mac all the more inevitable, if the Mac and the PC
> become increasingly similar. Migration from a Mac to a PC becomes a snap for
> most users. Indeed, to most people, there's hardly any difference any more.
>
> The problem for Apple is that they're trapped. Steve Jobs desperately needed
> Bill's money -- and the average Mac user wants to run Microsoft products. But
> having Microsoft be now, by far and away, the largest supplier of software
> titles for the Mac is much like being parasitized by a massive colony of
> intestinal worms. They're eating you from the inside out. If the user can't
> tell any effective difference between a Mac and a PC, Apple loses everything.
>
> Ted cites the differences in quality of design, reliability, and ease of use
> of the Mac over the PC. I'm not sure that there's that much difference any
> more.
>
> I've been a Microsoft BASIC user since 1975, a couple of years before
> Microsoft itself even existed (it was called Altair BASIC at the time), a time
> when BASIC was more than merely a language, it was also the operating system.
> And during that time, we were users of any number of early DOS'es, including
> North Star, MS-DOS, and CP/M.
>
> But in 1984, when we first saw the Mac, I was genuinely startled at the
> quality of the Macintosh -- and we were immediate and enthusiastic converts.
> We almost immediately dropped all of the PC-like devices, with the exception
> of those Altairs that were connected to the HP3000 as terminal-like PROM
> programmers (some years earlier, I had written 8080/8085 and 8086/8088 cross-
> assemblers on the HP3000 for easy compilation of code for the single-board
> computers we built during that period, downloading the code from the HP3000
> into the Altairs for transfer into PROMs). Otherwise, all of the early IBM PCs
> went to the dump, a perfectly appropriate repository for the quality of device
> they were. We designed industrial-grade computers during that time, and the
> IBM PCs were completely inadequate to the tasks at hand.
>
> I have always been amazed at the quality of the Macintosh's design. It is
> simple, clever, and elegant. And to a great degree, we've tried to imitate the
> Mac's philosophy in everything we've attempted to create for the HP3000. What
> I tell our customers now is what we're trying to do is to turn the HP3000 into
> a Macintosh. Simple, elegant, robust, and easy enough to use that an on-site
> data processing staff would not be necessary.
>
> As I was composing this note, a long-time friend wrote me privately:
>
> > If your business model keeps you from supporting a platform,
> >  fine -- just admit it, and do what you do best. But must you attack a
> >  platform you don't want to support?
>
> In that regard, my previous posting may have come across as an attack, but it
> certainly wasn't meant that way.
>
> I honestly don't believe that I have an axe to grind in this -- other than I
> believe that people who want to hang on to Macs in an HP3000 business
> environment are making a mistake that will only grow worse with time.
>
> My reasons for saying that are completely pragmatic. We still operate as an
> off-site system manager for a handful of HP3000-user small business
> organizations here in our local area, true Main Street businesses, the
> remnants of an HP3000 time-sharing service we began 22 years ago -- and people
> with whom I'm very pleased to continue our relationship. These people
> represent the model that I believe the future of the HP3000 to be. They are
> purely business people who have come to critically depend on their HP3000s,
> but they do nothing other than perform backups and change the paper in the
> printers. What little maintenance of their machines is necessary, we do
> remotely.
>
> Five to eight years ago, when their previous word processing capabilities
> (standalone word processors that we built) were coming to the ends of their
> respective lives, I strongly advocated that they all buy Macintoshes, and they
> did. That was no mistake at the time. Given the quality of the available
> alternatives, the Macintosh with its PostScript printing capabilities was the
> only possible choice. Anything else at the time would have been an inordinate
> waste of money.
>
> Just recently however, I had to revisit the problem with most of them again.
> The Macs were now coming to end of their useful lives. This time however, the
> most rational recommendation was to abandon the Macs -- something they wanted
> to do anyway -- and move over to PCs. They found that they increasingly needed
> to communicate with other companies and governmental agencies in formats that
> are far more easily supported on PCs than Macs. And they needed to run market-
> specific software that is only available on PCs, thus the PCs were showing up
> anyway.
>
> But much more importantly, the world had changed in the eight years. Windows
> had become good enough, reliable enough, and sufficiently easy to migrate to
> that there was no longer any reason not to migrate. To recommend anything
> else, no matter how enthusiastic I have been about Macs, would have been
> untruthful on my part. PC prices are lower than comparable Macs, there is a
> great deal more appropriate software available for them, but most importantly,
> there is the assurance of not being orphaned in the near-term future. When the
> situation has gotten to the point that no essential difference exists between
> the 95% solution and the 5%, there is no other rational choice from a business
> perspective.
>
> Can the Mac survive? Can the HP3000 survive? These are two completely
> different questions. The oldest rule in evolutionary ecology is called Gause's
> Exclusion Principle; it basically states that when two species, A & B, no
> matter their phylogenetic heritages (how they got there, regardless of their
> family trees/lines of descent), if they are direct competitors for some scarce
> resource, only one of them will survive, even if they started out evenly
> distributed.
>
> Niche partitioning -- and dominating your chosen niche, to the exclusion of
> all other competitors -- is critical to long-term survival. In that regard,
> the iMac is a very good idea, and Apple's only hope.
>
> The home market is not to be dismissed. If Apple can get the price low enough
> (and price is everything in this market) and they can make the iMac simple
> enough and easy enough to use as an internet terminal, and capture enough of
> the market mind set early enough, then they have a real chance to prosper. A
> statistic was published yesterday that 1/3 of all Americans over the age of 18
> now have access to the internet and use it in some significant way rather
> frequently. The internet, most especially the web, is going to increasingly
> become a necessary part of everyday life. Because of that, an inexpensive,
> information appliance that will allow you run a variety of high-quality basic
> software titles in addition to accessing the web has a real chance to succeed.
>
> The iMac isn't Larry Ellison's or Scott McNeely's idea of a network computer.
> It doesn't represent "the end of the operating system," as they would have it.
> Rather, an iMac-like device is an operating system-based machine that has to
> become once again simple, focussed and extraordinarily reliable for this
> market, not all that much different than a television set or a refrigerator --
> or the first Macs. No complexity. No fuss. It just works. But it has to be
> cheap.
>
> At the core of the problem lies this damnable tendency to want to be all
> things to all people. If a Mac is capable of running Linux, is it still a Mac?
> Or is it just another generic box?
>
> This discussion began on the subject of multiple operating systems. Let me
> coin a few terms: MOST (multiple operating system technologies), MORE
> (multiple operating regimes) and MISS (multiple information system
> structures). In computers that must compete in an increasingly competitive
> world, MORE is much less than MOST -- and MISS just barely makes the grade.
>
> The HP3000's POSIX is a MISS system, where two different file structures have
> been inserted on the same box. It adds some capability, but at the cost of
> some real confusion and a significant increase in complexity. But POSIX still
> isn't sufficient in and of itself. A substantial portion of the attributes
> that POSIX is aiming at emulating are missing. Thus, the inevitable tendency
> is to push on towards MORE, where multiple operating regimes, derived from
> different heritages, are blended together. But the problems of doing this are
> legion. Ultimately, as the accurate emulation of all of the various heritages
> becomes increasingly accurate, the solutions and the problems facing the
> system architects tend to become inextricably complex.
>
> As a brief example, Jeff Vance wrote just a few days ago:
>
> ========================================
>
> > One drawback is that if you issue a REPLY with no space between the
> > keyword and the pin, it's not trapped  (REPLY103,Y). Something to do
> > with numerics allowed in udc command names, which are parsed before regular
> > MPE commands.
>
> This is correct.  In retaining MPE V compatability, the MPE/iX CI follows
> the same rules in extracting the command name.  Namely:
>   - parse out the name using only alphanumerics (must start with alpha)
>   - see if it is a UDC
>     if yes the 1st byte past the end of the name begins the 1st arg.
>   - if not a UDC parse out the name using only alpha chars.
>   - see if it is a built-in CI command.
>     if yes the first byte past the end of the name begins the 1st arg.
>   (added to the MPE XL CI)
>   - if not parse out the name using MPE and POSIX filename rules.
>   - see if the name matches any file using HPPATH elements.
>     if yes the first byte past the name begins the 1st arg.
>
> Example command line:
>
> :abcd23.ef_g
>
> If a UDC named ABCD23 exists it will be invoked and passed ".ef_g" as the arg
> ElseIf a built-in command named ABCD exists it will be passed "23.ef_g" as
>        the first arg
> ElseIf an MPE named file matches ABCD23.EF[.acct] it will be invoked and
> passed
>        "_g" as the first arg
> ElseIf a POSIX-named file matches abcd23.ef_g is will be passed no arg.
>
> ========================================
>
> The current POSIX system on the HP3000 isn't yet a true MORE system. The UNIX
> system calls are missing, thus every UNIX-derived program has to be ported to
> the HP3000, no matter how simple it is. While we justifiably make heroes of
> the people who have taken the time to port this software over, it is an
> activity that should never have been necessary.
>
> With the advent of IA-64, there will inevitably arise the great notion that
> not only should the UNIX system calls be callable on the HP3000, under MPE's
> POSIX, but also that the NT 32-bit APIs should similarly be callable, with
> NT's file system similarly contributing to the MISS. While such a system would
> become a true MORE device, it would also become complex beyond all reason --
> and essentially non-evolvable due to its intrinsic complexity. Every
> modification has the chance of breaking multiple operation system regimes,
> thus nothing can be changed except with the greatest reluctance.
>
> Ultimately, the only way to break out of these problems is MOST, a design
> solution that is precisely the opposite of attempting to blend all of these
> operating systems together in a MORE-like fashion. Rather, under MOST, all of
> the operating systems are partitioned and firewalled into their separate and
> independent domains, with independent discs. In this design architecture,
> POSIX would be backed out of the HP3000 and UNIX and NT would be run as sub-
> universes. MPE would be made simple and reliable again. More importantly, MPE
> would also act as the monarch universe, because it is the reliable operating
> system. UNIX and NT would be free to crash as they feel necessary, but no harm
> would be done to MPE should they fail.
>
> The long-term benefits of MOST over MORE are significant. Most importantly,
> the sub-domained operating systems would be free to evolve as necessary in
> their relative competitive universes. Moreover, no porting of new operating
> system features or application programs would be necessary. UNIX under MPE
> would be the same as UNIX on an HP9000. The same would be equally true of NT.
>
> We're very rapidly approaching the point where this sort of decision is going
> to have to be made. The HP3000 is at the MISS point now -- and pushing on
> towards MORE. But NT is going to be the camel that is going to break the
> straw's back for this sort of design.
>
> Apple faces very much the same sort of problem. Their best hope is to make a
> very popular, very reliable, inexpensive internet terminal out of the Mac --
> and at some time in the very near future provide a MOST-like solution,
> allowing UNIX and NT solutions to be run as sub-domained operating systems,
> completely independently, so much so that only file read and write reachover
> capabilities are provided.
>
> Doing this will allow the Mac to be truly a Mac and distinguish itself for
> those things that it does extremely well. Apple has to do something to get the
> Mac out from underneath the 30,000-ton elephant, and this is probably their
> last chance.
>
> The HP3000 doesn't face the same intensity of problem. There is no other
> commercial database engine out there that is as reliable and valuable as is an
> HP3000. But that is no reason for complacency. The HP3000 has to be made more
> simple yet, not more complex, it has be made more reliable yet, not less, and
> anything that can be done to get the price of entry-level systems down will
> only do us all a great deal of good.
>
> MORE is a design path that invariably leads to complexity -- and as Ted says,
> no one needs a Mac that acts like a PC. MOST, in stark contrast, is a
> structure that promotes simplicity -- and yet allows complete competitiveness,
> if need be.
>
> Wirt Atmar
>

Jim Kramer /Lund Performance Solutions
Director of Research and Development
phone: (541) 926-3800  fax:   (541) 926-7723
email: [log in to unmask]    home:  [log in to unmask]
http://www.lund.com