>Here are some other possibilities and suggestions:
>
>1. Since only four users are involved, institutionalize the foregoing fix programmatically and call it a feature. Sorry.
>
>2. Use flush. It will force a write to disk. Problem is that your users PCs can create records faster than the disk can update, but VFP may think that the disk is being updated just because the flush was issued. VFP is actually writing to the network buffers...not to disk. You can "fix" this by putting a timed delay after flush. It works, but is not a clean solution.
>
>3. Add a flag in each record to indicate a user is editing. You do not mention file locking, but I am sure you do. Flags are an additional safeguard. They should be redundant, but may help if the locks are not working.
>
>4. Do not fix your indices (NOT INDEXES) by reindexing. Destroy and recreate them new. Corrupt CDXs can appear valid and accelerate problems.
>
>All this assumes the following:
> 1. Your network and PCs work properly,
> 2. You have no viruses, and
> 3. Your users are careful about turning off their PCs.


We are using Optimistic buffering in most cases. I gave you the wrong number of users. There are currently 40 active users.