I think you're right, that HP is not likely to be willing to sell the HP3000
business when they're tired of it.  When they tried to kill Transact a few
years back, I heard that they weren't willing to sell that either.
 
The many messages concerning HP3000 sites which are switching to UNIX
make me wonder if most companies that use the HP3000 are rolling in money
(cause & effect?).  Two of my clients seemed to be on the brink of joining
the lemmings recently.
 
One client has separate packages, customized in-house, for Manufacturing,
Financials, and Order Entry.  The management directive was to select one
integrated client-server package to handle all of these functions.  The
package
was eventually selected, and it requires UNIX.  Then somebody added up the
costs, so they could be put into the budget.  Estimated cost of purchase and
conversion: $4 to $5 million.  The HP3000's will stay.
 
Another client has one major application running on an HP3000.  It was
written
about 15 years ago, when the developers did not expect it to still be in use
when the turn of the century rolled around.  (I counted 52 separate date
fields,
and disc space was expensive back then.)  Modifications for the year 2000
and for other required changes were put off with the intention of switching
to
a new package, which would incorporate all the needed changes.  The packages
under consideration run on, you guessed it, UNIX.  After looking at the cost
of
purchase, conversion of data, and retraining, and perhaps some consideration
of reliability requirements, it looks like we'll be doing the modifications.
 
If HP does succeed in killing off the HP3000, as they seem to be hinting,
it looks like an expensive proposition for some companies.