Hi Jon,
Imagine the following senario:
Client locks a record on a file-server.
Client's PC is turned off.
IMO to correctly lock a record there should be no modifications to the dbf file.
VFP has to use server's OS for this. But more or less exaclty how it is done is shrouded in mistery to me.
Hope somebody could elluminate us on this issue.
Kamil.

>I've tried testing this theory - by creating a table with a single field and then doing a hex dump of the file with and without a RLOCK - but the results indicate that neither the table header nor the extra byte in the record are used for locking purposes...
>
>
>
>>>>>Does anyone know what VFP is doing at the OS level when an RLOCK is issued?
>>>>>
>>>>>I've been trying to find an answer to the question "Who has locked this record" in a multi-user NT based system so that I can monitor the locks.
>>>>
>>>>That may be a tough one to do in that way, you're looking more at SQL Server type of DB for that. However, I have done this sort of thing in vfp in a home-made way - when a user locks a record, you can write username and PK (and table if necessary) to a tracking-table on server, not difficult to do.
>>>Bruce,
>>>
>>>Thanks - I've already implemented a tracking table method with a wrapper to RLOCK and UNLOCK - but I am still interested in finding out how the OS handles lock as it would potentially enable implicit and explicit locks to be monitored
>>>
>>>Cheers
>>>
>>>
>>>Jon Nelson
>>>
>>
>>As far as I know there is a bit in each record that indicates whether it is locked or not. As you know, with a record that contains e.g. 99 bytes of data, 100 bytes are used for the record. The extra byte is used for indicating whether it is deleted or not, and I think also for the lock status.
>>
>>You could check out the description of the file formats that comes with VFP (at least, with the development edition). Maybe it's in there.
>>
>>There's also a byte or bit in the table's header to indicate whether the table is locked (I think).
>>
>>Marc