Hi Thomas,

Thanks for you reply. In my case I was first selecting records to the temp cursor, then applied GROUP BY on a result set. This temp cursor is not used anywhere. So, I made a quite logical assumption (I think so), that instead of pulling unneeded data over the network I would group by in the same SQL since I really need counts. I would assume a performance increase, though may be slight. However, I don't feel like I gained the expected performance boost. On the other hand, I didn't really measure times, so all of this is not scientifical...

Anyway, I already made the change and posted into VSS, so hopefully I didn't degradate performance <g>

>Hi Naomi,
>
>FWIW in your context: I do quite a lot of ad-hoc queries which have to get refined sometimes and/or put into app's. Most of the time I create temp cursors, leave them alive and work of them to create the next ones. Even when often working with data between 1 and 2 gig, I don't have perf issues often and can debug by logic more easily. But I have created a small framework with a property to decide on cleaning temp cursors only needed as oprocessing step. If i have to give out an app quickly writing a few result tables / excel files it is working without leaving them open.
>
>my 0.02 EUR
>
>thomas