We use a common base class to handle most Assembler functions without writing a separate DAO class for each destination. The only trouble we have in this design is using the API functions in AbstractAssembler which give no reference whatsoever to the Class of object to act on. One such function is getItem(Map). The Map only contains id information, with no reference to what kind of object to retreive. If the destination string was included, or a Class object, we could get any record in the system by writing only one function.
We have written our fill routines to pass in an object class name so one fill function could be written which dispatches the correct query logic. And, with create, update, delete, we have an object of the correct type passed in, so our persistence framework takes care of the rest.
This greatly reduces the workload of server side programming. We spend our time writing business logic instead of plumbing.... When designing FDS functions, please keep in mind that users of FDS wish to use frameworks in our applications, not brute force programming... After all, that's why we use Flex in the first place. Thank you.