Walter...

In reality I am trying to narrow down a set of functionality that a client tool must be able to implement.

For instance here are some items that I would consider essential...

Given a set of data... I need to...

Filter that set of data based on some criteria.
Sort that data based on some criteria.
Display that data in a UI
Edit that data via a form
Add new data via a form
Send that data back to the server for storage.
It must do these things in an efficient manner(efficient is a relative term which will become objecive we benchmark these items)

Those are some starting points. Another point of this exercise is to do it objectively. To me objectvely requires that we define statements like:\

>>limited proprietary
>> extensive DML
>> better data language environment

Also we need to examine statements like:

>>Cursors (and its indexes) are auto disk spanning, meaning that large resultsets can be >>flushed to disk transparantly, saving process memory so that otherprocesses will not >>starve from memory resources

This is another relative issue. And could be countered with something like… Disk access is orders of magnitude slower than memory or Windows does disk spanning natively via virtual memory access. We need to figure out how to quantify the advantage here.

Thanks
Rodman