>AFAIR you already handle the errors for the read cases by reading again. So only the insert/update cases need to be watertight - would this not reduce the number of locks needed ?

Not really, when the sites received simultaneous hits, even if I retry, that retry will be done among other simultaneous hits, so basically, they will bounce on each other.

I discussed that on phone with a Microsoft support VFP person yesterday and we all agree that this should be resolved as those kinds of workarounds are not acceptable.