James wrote:

> Good question Lester. As I answer my question why I need two
> ib_transactions, hope I might get an idea from you for my problem. I
> have a master detail setup in my application, wherein I used to check
> for possible duplicate records before saving the new record. As we know
> IBO has a feature called post retaining in master detail setup, if the
> dataset that will use to search for possible duplicate is attach to the
> same ib_transaction, the answer will be always yes since there is post
> retaining that is perform. Hope somebody have similiar situation with me
> and has a better solution for this =).

So the 'search query' is stand alone, with it's own transaction, and
called before/after each new record is created?

I don't see why you need to worry about the 'search query' seeing
anything of the 'new transaction' until it is ready to post, and then
the next 'cycle' the 'search query' will include the latest info.
The 'new transaction' should take care of checking any duplications it
is creating internally before the transaction is committed?

OR are you in a situation where several users could be adding the same
new stuff, so the checks need to be made when a transaction is completed?

The new question is "How are duplicates created, and could THAT be
handled better with a generator?" Once again a little knowledge is a
dangerous thing :) Without the whole picture suggestions for improvement
are difficult ;)

--
Lester Caine
-----------------------------
L.S.Caine Electronic Services