> My suggestion was that a query could always behave as though
> AutoFetchFirst was on when it has a MasterSource. I could easily put
> a patch for this together and submit it if nobody anticipates any
> side effects. If this condition is not sufficient, I could write
> code to check more prerequisites for ignoring AutoFetchFirst.

I cannot imagine any reason why this should cause problems, but does
not mean there aren't any :-(

If you can restrict your patch to TIB_BDataset (the buffered dataset
base) then I think that would be the safest option. I suspect that
unbuffered datasets (TIB_Dataset, TIB_Cursor etc) are more likely to
be adversely impacted by this enforced row selection because they are
used in code.

My personal preference for this would be to have the AutoFetchFirst
property *literally* kept/forced true when MasterSource is assigned. I
know some people prefer not to have properties interact in this way,
but to my mind it visually demonstrates the relationship between the
properties and avoids someone looking only at the AutoFetchFirst
property and wondering why it is not working.

--
Geoff Worboys - TeamIBO
Telesis Computing