Barbara,

>"Flush" sounds like it SHOULD work, but we found it had absolutely no effect on a network application. It can only send data to the network buffers, and then "other forces" take over from VFP.

See my reply to Eryk, why flush does not work.

In a network application, flush has only effect if the workstation holds writebuffers. This is only the case when :
a. You've opened the file exclusively before making changes.
b. You're the only one who has openened the table on a NT network where opportunistic locking is enabled (default). This opportunistic locking allows the workstation to hold writebuffers unlit another workstation wants to open the same file.

In short, FLUSH forces the writebuffers to flush, but does nothing to ensure the readbuffers to be refreshed. In this case, I think the problem lies in the read buffers. The only (reliable) way i've found to refresh the read buffers is by SET REFRESH, or placing a record or file lock (Either implicit or explicit).


Walter,


>>Walter,
>>
>>What about FLUSH command?.
>>No one has mentioned it yet. It's supposed to save table and index modifications to disk.
>>Is locking of a table or a record more reliable?
>>
>>Best regards,
>>Eryk