We share lots of tables over the network. The network drive is Novell 4.11, with some NT domain controllers thrown in. Our clients are NT 4.0 workstation sp6. The users use VFP6 runtime. My tests were on VFP 8.

We have a one record table with a boolean field. I flip the boolean in the table, but it doesnt get refreshed.

Buffering is not involved.
SET REFRESH is set to 0,5

Editing the table will force a read off the network, as will GOTO 1. Is this really required? Shouldnt VFP reload table in a record from the network every 5 seconds? The data seems get written fairly rapidly.

An RLOCK or USE will force a read off the disk, but dont suit our needs here.

It appears that changes to fields do not get updated until the cursor leaves the field. Do changes to a record get written to disk only after the record pointer changes?

Over the weekend we've had some strange behavior. We flipped the flag Friday afternoon, and it took effect around noon on Saturday. I expected a five second lag. I could understand an indefinate lag. But I don't understand a 12 hour lag.

In a similar test a flipped the flag and waited for another instance of VFP to recognize it. It took many minutes. Local drives also appear to have this behavior.

It seems strange that VFP should behave this way, as I've never noticed it before. The whole locking and updating mechanism has always been almost flawless once you understand buffering and set refresh.

Is this expected behavior? Or a strange novell glitch?

Thanks,

-hans