Various people responded with helpful pointers.
Thanks.
I'll attempt to respond to each in this email to avoid clogging the list.

To all who sent links, be they Robelle, Wiki, Hp, etc, thank you very much.
I've printed out the TurboImage FAQ for my evening read, and saved the rest
for 
reference.

> Also ... What you want to do is best done through Powerhouse and that has
to 
> be done on the HP3000

I might end up there, but if possible I'll avoid it.  Our HP-3000 is very
old 
and has a lot of cruft, runs inventory, reporting, and lots of other core
business functions. I might end up doing twice a day (or more) dumps, and
that will
probably have a significant impact.  

I'll need to test and measure before making the decision, but I need to know
what my options are.



> You wouldn't want to do this unless you knew what data you needed moved 
> elsewhere , would you? 

Everything.  Every record in every inventory, receiving, pick, order, (the
list goes on and on).  But the system will be core to our business for
probably 
another year while a 3rd party package gets installed and we start using
it.  Until then, we need lots of new reporting and QC code written against
the extracted data, which will probably end up loaded into a SQL database 
(MS/SQL - acch phhtt!)

> My suggestion is to use powerhouse to extract the data you need into flat
files, 
> in a format of your choosing (via quiz specs), and ftp that elsewhere,
probably 
> in ascii would be easiest. 

Probably easier, but until tested, I don't want to commit to a particular
solution
if I can't be sure it won't impact the current usage.

> Since you seem to be able to get into powerhouse already , and know some
commands, 
> you may be halfway there.....Unless you are talking HUGE volumes,
Powerhouse can 
> be a pretty efficient method of doing this.

I've got essentially unlimited CPU/IO cycles on other boxes.  I'm severely 
constrained on this one.

> Depending on the destination box, you may run into trouble with floating 
> point values...

Yup - major wildcard.  I was actually expecting to do some low level
bit-twiddling,
assuming I can find docs that clearly explain the format.  I'm bringing the
data 
into Intel 32 bit Linux.  My code is typically Perl by default, and it has a
decent set of conversion routines in the unpack statement, but all bets are
off
until I see the data.

> the short answer -- is you can't directly extract data from an image 
> database without going through <something> that provides an api layer.

Acck!  You said "can't".  I refuse to accept that (in my ignorance) (yet).

> - suprtool.  do a listf to see if suprtool.pub.robelle exists on your
system(s).  
> if so, run (don't walk) to:

Yes, we do have suprtool.  My boss (who is the HP person) gave me the
manual,
But I wanted to spend at least a couple of days wandering around, doing
tutorials,
at least figuring out how to do directory listings and the print command to
dump
files before I went to the next level of a tools that masks the low level
stuff.

> for all the above options, you'll probably need to consider placing the 
> extract process in a job stream (kinda like a script) and streaming the
job 
> in off-hours.  (the job can also include ftp'ing the extracted data 
> <someplace else>)

We already have a bunch of scripts (streams) that due that piecemeal right
now.  
The problem is that we seem to come across conversion (business logic)
issues
that can only be resolved by going back to the original data, again and
again,
and the tools on HP are very limited in comparison with what we have
elsewhere.
So if possible, I want it all, in an ongoing basis.

> there are options beyond what i've listed....  they all involve
money....and 
> may great improve the quality of your life....for a price.

I've already traded in the QOL issue for a price.  Now I gotta cover the
bill.
Note:  That's not to say I WON'T end up paying for expertise or a tool to
help, just that my boss has an inflated expectation of my abilities in
systems
I've never seen before, so I gotta at least try to deal with it.

> just be glad you've found this list.  we've all done what you're doing
1000s 
> of times.       - d

Thanks




* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *