> How about opening a new cursor for nulls when all the non null values are
> exhausted?
> (Of course only as long as it's not a "not null" field).

What is stopping me is the amount of complexity that then fans out on the
rest of the operations that the dataset carries on with the server when
doing locate and filtering for example. It's possible, just a lot of work to
keept it clean to the end user. For now I just need to document that it
isn't supported.

I'll finish this up in a future release.

> BTW, does HDR work with multi field orderlinks?
> Because sometimes when I have a nullable column or one with many

duplicates

> I also sort by a second field.

It does work functionality-wise but for some reason there are performance
implications that the server doesn't know how to handle very well and
performance can seriously degrade.


Jason Wharton
http://www.ibobjects.com