>Hi all,
>
>A preferences and practices question.
>
>I have a VFP6 app that I am implementing and maintaining (written by others) that in the course of the user editing records, the developer does a lot of copying the records being updated out to a temp file, processing the edits, appending the records back to the tables, then deleting the original records.
>
>It seems that this would be a performance hit when the edits encompass a lot of records (this is a work order and inventory app), as well as requiring a pack every once in a while. It just seems easier to edit the record directly, but is this possible in a multi-user network app? It is a procedural rather than OOP style app. Looks more FPW2.6 than VFP. I want to know is this the best/preferred way to do this?
>
>TIA

Well, it would not be my design by choice. There are too many things going on and after all what is record buffering for? I have written several applications concerned with work order and inventory systems. These are all multi user systems and they can get interesting when it comes to committing an order with stock on hand vs backordering, etc. You could consider a different data model than the one you described but you have to weigh everything.

1. Budget.
2. If it works at an acceptable level – leave it alone!
3. If you have the need, budget, time and managements approval – redesign it.

I never change anyone’s code without a real need to do so. Nothing is perfect and no ten programmers will agree on one way to do anything. At least not the ten I know! :)

Tom