Reaction to VFP3 was rather negative - terrible performance and a rather unhealthy list of bugs.

Certainly the impression I got about VFP from most of those with whom I dealt was very positive and (believe it or not) I even formed my own opinion during the Beta and subsequently.

There was an undertone at some of the conferences back then - subtle dancing around the VFP 3 deficiencies. I know many who wished that MS would have simply taken 2.6 for Windows and natively incorporated stuff like GenScrnX and TabX, etc. Some commented that VFP3 was not totally unlike Ashton Tate's dBase IV.

People who had done C/S with FP2.6 would have had no difficulty whatsoever seeing the improvements in VFP. ;-) Jeff Winchell and I had some fairly robust debates on the differences between dBase and Fox at that time with VFP focused on parameterized queries with automatic change tracking while dBase was off in a different non-parameterized direction. Even today I can think of pages of VFP items that offered huge improvement for developer and customer. So I'm not sure what your contacts might have been talking about there.

It took MS until August of 1996 to release a much better VFP 5.

OK.

So there's a "silent majority" of folks who tackled many of these problems and moved on.

Perhaps that'll be the same "silent majority" who apparently don't see any need for change tracking. ;-) Yet MS built not only snapshot but also virtual change tracking into POCO. If people want to "move on" they need to be thinking about these things.