>>Maybe relying on SQL is hurting me and I don't understand why. Can you explain what I'm missing?
>
>This question is obviously out of my reach. I don't know. As you understand I have heavy VFP and spotty SQL-Server experience. Actually, I was requested more to write VB-SQL, and only afterwards I converted some to VFP-SQL. My opinion (very limited and personal) is that C/S concept is slightly outdated and outpaced by WEB (unfortunately, my experience in WEB is also spotty and never included VFP tools), while local data and native access is still alive (i don't know how long).
No problem here; you've not seen the need to integrate VFP and backend systems to date, and no amount of proselytizing about how good it is where there are compelling reasons will change your perspective. I'd guess that this will change over time.
I'm really lost by how Web-centric applications outdate and oinvalidate client/server database technology. From my POV, the Web application is even more narrowly focused on small, user-defined and unpredictable data sets, and backends, with stronger security and the ability to scale to meet unexpected demands better than VFP, fit into the environment even better. My first detailed involvement with backends came from mid-tier business logic designs for fairly large data requirements, where VFP handled the logic well but didn't scale the way the client needed it to because of transaction volume. Web apps seem to penalize you a great deal for moving and buffering large amounts of data, both in terms of UI and server resource requirements.
>One more thing I feel compelled to say. Actually, OO-approach is a key that moved what called procedural and code-intensive in 2x days to much less coding ground.
I agree that OO is good, independently of whether you rely on SQL or xBASE for data handling. It encourages a different perspective of division of labor and taks responsibility. SQL's approach to stating the requirements of a solution in detail rather than the means of achieving the solution in detail are OO at least in that they delegate the exact process mechanism to the data engine. If the engine changes it's basic behavior, I'm less likely to suffer side-effects, since what I want hasn't changed. The best process of finding that solution might.