Grigore,

I don't know if you are talking to me or Jim Saunders (but I do know it isn't Yuri < s >). I'll assume me...

I have no idea why you would call this an "undocumented feature" unless you are being funny, and your remaining words suggest that is not so.

You should always assume that your code WILL work in future releases UNLESS specifically told otherwise, and then it would be restricted to very very specific items.
This was a lesson "taught" to IBM in the mid-1960s and they almost lost their business because they didn't operate with that policy. They then had to promise that, FOREVER MORE, all their systems would be backward compatible when a prior system existed.

In the case reported here Jim Saunders (Microsoft) has acknowledged it as a bug and has reported that it is already FIXED (in later beta versions).

Most of this thread was arguing IF this was a bug or not. We now know that it was a bug!!!!!!!!!!!!

cheers


>Jim,
>
>The code shown uses and 'undocumented feature', which was eliminated in VFP8. In the same way, let's suppose the VFP 9 won't reselect automatically the initial table in a SCAN...ENDSCAN loop. In this case, almost all my code will crash - I don't reselect the table before ENDSCAN.
>
>Question: Would that behaviour be a bug? Wouldn't? Not reselecting the original table is an 'incorect code implementation'? In my oppinion, if it works, it is correct, but I won't assume a particular code will work in future versions.
>
>>Sorry, but you did not add a new argument. Still from my point of view (I know, that field name should not have spaces at the end) it is incorrect code implementation. If stored in the structure table, I would retrive field names with alltrim().