Currently our app uses multi-table views to load our grids. We generate a statement ( "CREATE SQL VIEW AS ..." ) at runtime based on columns the user selects and then we open the view and set properties accordingly. Columns could be standard table columns or calculated fields requiring a function call ".. AS ... " in the query. We are doing some restructuring of the code to better allow us to connect to SQL in the not to distant future.

We want to use Cursor Adapters. I have been playing around and see I have many options. Should I create the same query string and load it into the Adapter object, setting all the appropriate properties and execute CursorFill() or should I go ahead and run the statement as a create view and save the definition and then open it with a use command and then attach it to an object, keeping in mind the view definition will change when the user chooses to add/remove columns to the grid and the whole rebuild process is started again. I seem to think that letting the Adapter do all the work is better, but any other arguments for/against?