Unlimited size result sets

["Buck" <buck.calabro@xxxxxxxxxxxx>]

Yes it's me again.  My Java person insists that she wants me to return 500k+
rows of data to her in a result set.  Those of us on the iSeries side think
this is not a good idea.  She says that this is how things work in the real
world (as opposed to a page at a time.)

In the interest of gathering benchmarking data, we'd like to try.  The catch
is the number of occurrences for a multiple occurrence DS.  The obvious
workaround is to base the DS on a pointer, which works great up to 16 megs.
But my result set is going to be much larger than that.  I can certainly
ALLOC more memory and use another DS, but I don't see any way to guarantee
contiguous memory, and without that I can't guarantee that the result set
will be right.
  --buck