OK, my turn....

I agree that there are two-no, make that three...oops, it's four-major
issues involved:

1) Hardware - Since the focus is on a low-cost-but-useable development
machine, a minimum-configuration box won't do. In particular, running the
OS, TurboIMAGE (with user logging), ALLBASE/SQL, the development tools,
and the application being developed (under the debugger), all over the
network, is not something I want to try to do in 32MB, or even 64MB, of
RAM. My Pentium Pro/NT box at home has 64MB-it isn't enough. The dual
P133/NT machine at work has 112MB-it isn't enough. Memory is cheap. I
think we're talking about 128MB or 256MB as the base.

With respect to the actual box, I think that the 918 box is fine-which is
good, because there is really no other candidate at the moment. Dreaming
about a portable 9xy is fun, but it isn't going to happen any time soon,
and may not happen ever. In the absence of an improbable sequence of
changes in the environment, the box we're talking about has a likely
worldwide market of a few thousand units. That isn't a few thousand units
the first year or two-it's a few thousand units ever sold. Developing a
"new from scratch" box for that size market isn't going to happen-it's
about two orders of magnitude too small. Porting MPE/iX to some of the
more portable HP/UX workstation hardware might get done, and we can
certainly hope for it, but I won't hold my breath. We've never had a 918
around here, but it looks to be a reasonable size and to weigh about
70-80 pounds. That's "portable" enough for me-I can take it out and put
it in the back seat of the car (properly strapped in, of course-"Seat
belts save drives") if I need to take it somewhere.

So, the box I see is something like this (swiping the first part of
Michael Gueterman's list):
918 (probably limited to the 918LX box) or more portable box TBD
128 or 256MB RAM
4GB HD (one drive, so you can add another later if needed) (disk space is
also cheap)
CD ROM
DDS-2

 By the way, I don't particularly care if the box is a "used" machine
that HP got back when some site traded in their 9x8 box for something
larger. As long as HP warrants and supports the hardware, "experienced"
electron carriers are fine by me, since electrons are relatively low-wear
cargo.

2) License - 4 user. A developer will almost always have two sessions
running, and our limited experiences with MPE/iX reaching the hard user
limit have been less than pleasant. With a 4-user license, I can set
":LIMIT something,3" and protect myself from reaching the hard limit.

3) Software - every developer-oriented product that HP sells. This
includes TurboIMAGE, ALLBASE/SQL, KSAM/3000, all compilers (which
includes Transact/iX and the other "Rapid" products!), VPlus, XDB,
Glance/iX, etc.. Basically, if you go through the "MPE/iX Documentation
Guide", all products covered in the following sections:
  3 - System Management Core
  4 - System Management Core Plus
  8 - ALLBASE/SQL
  12 - Languages
  14 - Developer's Kit
  16 - Programmer Productivity Tools
  17 - Programming Core
  18 - Programming Core Plus
  20 - System Di-just kidding

This does not mean to actually ship these manuals; they're just a
convenient way of building the software list, though with a lot of
redundancy. The software and all documentation would be on CD ROM along
with the OS.

4) Support - Here's the tricky part. Or rather 3 parts:
a) OS and product updates: these are critical, and they should be as
automatic as possible. The easiest/cheapest way I see to get this is a
standard software release on CD ROM every <insert value here-3-4
months?>. You update from the CD, and everything that's been changed is
brought up to date. The sale price would include the first year of
updates; after that, the update subscription would be some nominal
(USD1000 - USD2000?) per year.
b) Hardware support: Gateway, Dell, etc. will sell you a PC for USD2000
with one year of on-site hardware replacement support included. Extending
the support to 3 years typically costs not more then USD250. Given HP's
vaunted hardware reliability, I would certainly expect nothing less. Now,
this does not mean 4-hour or even next-day response; the bundled
hardware-replacement the above-mentioned companies include is more on the
order of a few days to a week or so. That's fine by me; if the machine is
so critical to my operation that I can't afford to have it down longer
than a day or three, I'll either buy two of them or pay extra for the
higher level of response.
c) OS and product software support: I'd like to see a "pay as you go"
plan, with the proviso that a call made due to a confirmed defect in the
product, OS, *or documentation* is a no-cost call. HP should take a page
from the other industry players that already do this. If someone wants a
higher level of hand-holding, it should be available as an extra-cost
option.

So much for configuration; now let's talk price.

In a word, damfino. I won't pretend to have a reasonable "hard limit"
price. However, there are some other numbers that may be of interest:
1) A (very) high-end development machine for doing PC- or Mac-targeted
development can be built for USD5000 retail. In the PC world, we're
talking about a dual Pentium Pro/Pentium II (not yet, of course) with
128MB RAM, several GB disk, Windows NT, high-quality (but not
3D-graphic-workstation level) video, networking, and DDS backup. I'd
expect to pay more for our putative 908 box, but not a lot more, since HP
shouldn't be expecting to make a tremendous profit on the hardware.
2) The "base" 918LX on HP's Web site is USD14000. That's for an 8-user
system, and I'm sure that it includes a) quite a bit for the 8-user
license, and b) reasonable profit.

So, we're probably talking about a 4-digit price that starts with a 7 or
an 8, but maybe a 6.

Why should HP sell such a box?
1) *Not* to make a big profit on the hardware. Remember, the target
market for these boxes is small development shops, one- or two-person
consulting firms, and the like. IOW, *people who would not otherwise buy
any MPE box of any kind*. Obviously, HP has to break even on the
hardware, but the pig picture is that HP is making zero profit on these
people now, so anything greater than zero is a bonus.
2) *Not* to make a big profit on the software. The incremental cost to
add another compiler or tool to the proposed "everything" CD is very
close to zero. So it makes sense to load on anything and everything that
might help the developer produce something that s/he can sell to an MPE
production site. Again, *this software is being sold largely to people
who currently have zero MPE software, and will never probably have any
without this box*.
3) To provide a very low-level entry point for small offices and the
like. Some posts to this thread have talked about rules/limits to ensure
that only developers have access to these boxes. Why? If a 7-person
dentist's office buys one of these machines for its
accounting/inventory/whatever, so what? On the contrary, I'd say "yes,
please!":
a) Any shop that comes to depend on the box for its office operations is
likely to want higher levels of support, more users, more CPU horsepower,
etc. So HP gets a built-in upgrade to a "real" customer.
b) If the putative "pizza box" machine appears, the above-mentioned
built-in upgrade includes a hardware upgrade to a bigger box.
c) By increasing its market presence, the HP3000 becomes more "standard",
more well-known, more talked-about, etc., etc. Something about a snowball
rolling downhill....
4) Oh, by the way, the major thrust behind this discussion: by increasing
the number of developers and consultants using and writing software for
the HP3000, more solutions become available that include the HP3000, more
customers can consider an HP3000, more people buy HP3000s....

Actually, my terminology is wrong: HP wouldn't be "selling" this box;
they'd be making a zero-cost, near-zero-profit "investment" in the future
of the HP3000.

Steve