> You don't need to be in too big of a hurry to get rid of your TIBOTable
> components because TIBOTable automatically makes use of horizontal dataset
> refinement if you have supporting indexes in the database.

I would also like to add, horizontal dataset refinement received almost a
complete rewrite in version 4.9.14. It was actually an agenda item for IBO
5.x that I decided to go ahead and backport since I had done everything else
and I had some issues I needed to fix in 4.9 anyway.

There were previously some gross inefficiencies in the way things were
handled. I only used two cursors that had to be re-prepared when you went
from the eof/bof portions to the middle of the dataset. Either that or the
input parameters were fixed for both zones and I had to use potentially
faulty min and max values to emulate getting the eof and bof segments of the
dataset. This had complications for collations, floating points, datetime,
etc. that were awkward and potentially problematic for some people's
applications.

Now, there are 4 internal cursors instead of two. All of the awkwardness of
having to come up with min and max parameter values and re-preparing cursors
is done away. It is now very clean and failsafe in how the dataset is
handled (nulls are still an issue) and it is efficient in that once the
query is prepared you will not have to re-prepare it just because you are
navigating through different zones of your dataset.

Regards,
Jason LeRoy Wharton
www.ibobjects.com