I didn't think there was much trouble with data handling in Delphi. Very similar to what VB apps do.

IMHO much of the issue is that not everybody was using C/S in 1994-1995, and those who were had become accustomed to programmatic SPT, just like in VB whether the data was in dbfs or Oracle. The shift to ADO (from DAO) and subsequently to ADO.NET and "SP is best practice" would have felt like no big deal. The RV essentially was a wrapper over SPT that could have been replicated in any of those other tools, and in fact was by some 3rd-party providers who called it a "framework" and charged $$$$ ;-)

Borland had its own ideas about data at the time. dBase had a new C/S mechanism that allowed you effectively to "open" a C/S table and without loading the whole table, browse through it like a dbf, looking for records and filtering on the fly and lots of other stuff with pages being pulled as required and saved etc behind the scenes. Winchell was a great proponent of that, but I found it too expensive and unpredictable when the only step it really avoided was initial parameterized querying for a specific dataset. Borland also had a remarkable pseudo-view that turned out to be the equivalent of a series of "Set Relation" between tables, iow didn't scale. At the time I decided that I'd go with VFP whose underlying injection-proof SPT felt less proprietary.