>Ed,
>
>This particular bug seems particularly frightening. Even after reading through a couple of long threads, I'm still not sure exactly when the problem occurs.
>
>Is the problem with both W2K professional AND server (assuming you are working locally on the machine)? If it only happens with local drives, can you create a network share and use it locally (i.e., map to your own shared drive) in order for Win2K to treat it like a network drive?
>
>Are FORM / REPORT files the only files corrupted?

I've only had SCT files corrupted; other people have reported FRT problems, and if that's the case, I'd probably expect some similar problems with VCTs - something involving the handling of internal system metadata files perhaps. It's not a matter of partial corruption, or a buffer not being written - it looks like the wrong data is written, and you're safe as long as the internal memo field buffer is held in cache - but the data remains vulnerable long after the file is closed and often even if VFP has been exited.

>
>Does it only happen to those VFP files residing on the IDE media (not copies of those files that can be made onto removable media before the system is shut down)?
>

I've only seen it happen on IDE media; at least one other person reports problems with a SCSI system that from my tests should be reliable - I use essentially the same controller and drive and driver, but I've never had a problem with the SCSI vbolumes I use. I've not seen any problem across a network drive unless there was a problem at the server end.

>Does it only happen if you exit VFP and then shut the system within a short period of time?... or is the cached data NEVER written to disk no matter how long you wait between the end of a VFP session and system shutdown?
>

It seems to happen if you've modified a Form and subsequently shut down Win2K abnormally after that, even having exited from VFP. I've had no problems with SCSI drives with WCE mode Disabled.

>Does it also occur if you hibernate your system (which I do all the time), and then start it back up?
>

I don't know - I don't hibernate.

>I'm really astounded that this delayed-write bug has only been reported with VFP, especially since Win2K is supposed to be a mission critical OS. One post mentioned that VFP apparently doesn't update the file date when an SCX file is saved... but how does VFP control that? (I have not seen this behavior). Isn't file date maintenance an OS file-system responsibility?
>

If you don't bypass the OS, yes - VFP has been known to bypass things on occasion.