This is the most confusing part of my team's development effort because
there are so many different options.  I need to integrate Turbine with a
legacy database and application.  We use Ingres RDBMS, and our keys are
stored in a table called ii_sequence_values.  The legacy application has
been managing this table, which looks something like this:

ii_sequence_values
------------------
table
column
value

This looks a lot like id_table, but it is used to generate sequences
that also serve as PK's.  You may have noticed in the other thread I am
trying figure out how to use IDBroker to make it generate sequences as
opposed to contiguous "chunks" of id's.. I realize this is somewhat of a
hack (but I checked and it does work!), so what are the other options?

So far I've seen the following options:

1) Just manage the sequence at application level.. I guess we could use
a Peer/OM to grab keys and update the ii_sequence_values table; then we
have to invoke this every place we insert an object!  So much for
automating it with Turbine.
2) Use a system-generated key for this field.  What does this involve?
Merely specifying that Turbine use the system (native) method of
generating fields?   Then we get rid of this table of keys.
3) Somehow get the IDBroker to generate sequences (done).
4) if we can get #3 to work, then could we modify IDBroker to use the
ii_sequence_values table?  This would allow existing code and new code
to INSERT INTO the same tables seamlessly.

Which is the best method?  The least painful (little or no data
migration)?

Thanks,
Nate

--
Nate Reed
Physical Oceanography Distributed Active Archive Center
Jet Propulsion Laboratory (Raytheon)
[EMAIL PROTECTED]
(626) 744-5528
(626) 744-5506



--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to