Wirt Atmar ([log in to unmask]) wrote:
> Richard Gambrell writes:
> >Could someone comment on the date functions, if any, in
> >Posix? I am very concerned that MPE and CI functions "reinvent
> >the wheel." We should use Posix functions when and where
> >they exist, in my opinion.
> Let me politely but vigorously disagree with Richard. Richard has been
> posting to this list often -- and if you do that, sooner or later somebody
> is inevitably going to vigorously disagree with something you said. And
> because it is inevitable, Richard should take no offense at what I am
> about to say, but I could not disagree more.
And, in that spirit, allow me to strongly disagree with a couple of your
points:
> Posix is not Unix. It can never be Unix. It can only be a bundle of
> significant compromises, a poor substitute, as all emulators are, for the
> original. And the closer the HP3000 becomes to being Unix-like, the more
> danger it is in.
You're right: POSIX is not UNIX. However, UNIX is not necessarily UNIX,
either (UX, Ultrix, AIX, SYSV, BSD, whatever). POSIX is supposed to be a
"least common denominator" standard that will allow (at least some degree
of) portability between OS platforms -- non-UNIX included.
> If the HP3000 is to survive and prosper, it must be thought of and
> designed as if it were a Macintosh -- simple, easy to use, and
> exceptionally reliable.
I think the Mac is the *last* thing to emulate what with their
"proprietary everything" architecture.
We're mostly Mac here (150 Macs / 15 Windows and NT) and I use both Mac
and Intel platforms daily and I can't say that Mac is any more "reliable"
than non-Mac PC's. Easier to set-up, definitely. Easier to use,
debatably (I don't like being forced to use a rodent) Easier to corrupt,
yes. When they do break, they cost more to fix.
Look at what software is available for Mac -vs- Windows. I would *love*
to be able to develop client/server apps on the Mac which access our 3000.
Why isn't there software for the Mac out there to let us do that? Lack
of installed base due to "proprietary everything".
Look, I'll be the first to agree that DOS and the arcane x86 Intel
limitations we still have to live with stink and the Macintosh and 680x0
architecture is much more elegant. But if elegance alone were enough,
then there would be no need for SoftWindows or Apple's PC-on-a-card for
the Mac, no?
In short, for the 3000 to survive, it must be able to run "industry
standard" software as painlessly as possible and as efficiently as
possible. This, regretably, means **IX. Whether it's elegant or not, it
*is* the industry standard.
Since HP's implementation of POSIX sits on top of a highly-tuned kernel
BESIDE (instead of on top of) MPE, my opinion is that HP is trying to give
us the best of both worlds and allows me to (more easily) defend having a
"proprietary" platform like the HP-3000 in our shop.
I think POSIX will be a key to the HP-3000's survival. Not its demise.
> All significant enhancements must be done to the basic structure of MPE,
> never solely to the Posix shell. Nor must the enhancements be done in a
> hurry or carelessly.
[rest deleted]
I'll agree with you here (and with a lot of what you go on to say). I
don't see how the simple presence of POSIX - especially the way HP has
implemented it - would necessarily stand in the way of the above.
Though we haven't received 5.0 yet, from what I've seen (other than the
additional disk space requirements) unless you actually *use* POSIX, you
wouldn't even know its there. Or did I miss something important?
Jeff Lindberg, Data Processing Coordinator
---
Mississippi Bend Area Education Agency ([log in to unmask])
729 21st Street Bettendorf, IA 52722
"I am Homer of Borg. Prepare to be assimilat... Oooo, Donuts!"
|
