>>>>
>>>>This was a really good technique in 2.6. But in VFP you don't know if the file returned by DBF() really exist on disk or is just in the memory. VFP doesn't guaranties the file really exist and my conclusion is (maybe wrong) that VFP writes the cursor to disk only when it doesn't have enough memory. (DBF() always returns a valid file name.)
>>>
>>>This is from the VFP Help:
>>>
>>>"Including NOFILTER can reduce query performance because a temporary table is created on disk. The temporary table is deleted from disk when the cursor is closed."
>>>
>>>Without the NOFILTER clause, the assumption of VFP using memory for the cursor would seem to be true.
>>
>>I know this. But there's a bug in VFP and NOFILTER doesn't force the file to be created on disk.
>>
>>Vlad
>
>Check out the thread "Phantom Cursor Filename" starting 7/12/97 (or 12/7/97 for non USA) in which a bunch of us ran various scenarios building cursors with dummy fields, nofilter, etc. NOTHING would force an actual on-the-disk, visible-in-MSExplorer file reliably. If there was enough memory the cursor stayed in memory despite what Help says and despite dbf() giving a file name.
>
>Barbara (original author of the PCF thread)

This is thread I refered to. But Mark Frank can't replicate this behavior. Maybe he has the solution for this bug!

Vlad