Some VFP people value the feature. You say not all- but I think you'll find that many/most developers customarily use/used local datasets for various purposes without being concerned that the resulting cursor might overload memory and crash their or their customer's PC. Customarily? I don't think so. Maybe in an internal/desktop environment, but as you move into distributed systems, I think you'll see less and less. I've worked in VFP shops where the goal was to avoid massive amounts of data on the desktop.
This is completely different from normal incremental chances that occur in IT. I don't think so. The experience of moving from A to B (be it network systems, service providers, databases, source code systems, etc), where B is both a subset and superset of A, is common enough that what I described is relevant.
