>>>>I am using local views and optimistic table buffering as a rule. When updating the views with 'send sql updates' = .t. (vfp5.0a) the updates are normally sent to the base tables automatically. I keep seeing mention of double buffering side effects but no real info. What are the considerations of double buffering and what do you do to make your the updates reliable? Do I need to specifically do tableupdates on the base tables to be safe?
>>>
>>>If you use buffered views, then base tables can be kept non-buffered.
>>
>>Thanks for your reply. I've done a little research and experimenting trying to understand the basics... When you have a buffered view as above and you don't add the base tables to the data environment, they are implicitly opened with no buffering. Its ok to send the updates to the base tables without buffering as long as there is no chance of network contention, (as in appending a new record). But if the possiblity for contention exists then the strategy should be to have the base tables buffered, either with cursorsetprop or by putting them in the DE, and do a tableupdate on both the view and then the base table all wrapped in a transaction.
>>
>>Does this sound right? Am I missing any gotchas?
>>
>>thanx again..Terry
>
>Buffered view itself is a tool to handle multi-user contentions, so base tables can be left non-buffered.

I get it now. When the view sends the sql update, it checks for changes to the base table and if changes have been made(to updated fields) the tableupdate of the view fails. Right?