>>>Hah? VFP8 implementation is consistent with ANSI standard. Why would they do it dufferently?
>>
>>No, I don't think ANSI has a "SET ENGINEBEHAVIOR". It is the VFP8 IMPLEMENTATION that I would rather have seen done differently.
>>My preference would have been along the lines of reporting the potential problem of ANSI comaptibility in a SYS(3054) message or such BUT otherwise leaving either as valid within VFP. That way no code would have been 'broken' and implementation of VFP8 could happen unimpeded with plans to later address the ANSI issues.
>
>Jim,
>
>As Tamar and I demonstrated, the problem wasn't ANSI incompability in VFP7 but possibility of incorrect result returned by queries.

Sergey,

It was YOU who said "VFP8 implementation is consistent with ANSI standard.", not me!
I then explained that they could have retained the ANSI compatability AND the fixes with a different implementation that didn't 'break' existing code and that would have been my preference. But, as I said to Tamar, we get what we get.
To be clearer: They could have fixed the bugs and left all else alone. They didn't, opting instead to add conformance to ANSI as well. I feel that they could have chosen to provide an ANSI compatibility WARNING in SYS(3054) output and let BOTH ANSI and old (non-ANSI) syntax work.
Tamar has explained that various options were discussed and that the chosen methond has its benefits. While I disagree, I do accept that.
But hopefully this discussion will cause additional reflection the next time that a similar choice is called for.