Re: Speed in Reading

On 07-Jun-2010 16:20, Peter Connell wrote:

But it ["always running SETOBJACC prior to accessing the object"]
would still do the trick.

Not necessarily, because as was already suggested, the previously loaded [even the most recently loaded] data into the memory may already have been paged out, even before the SETOBJACC completes. And even more likely, to have been paged out before the request to process the data actually starts... even if accessed by the next CL statement, the request may still have to fault the data into memory.

Attempting to optimize the effect by issuing SETOBJACC repeatedly into a non-dedicated pool, "always running SETOBJACC prior to accessing the object", is little different than optimizing each trip to a take-out dining place by calling ahead with your food order. While very often you will arrive to your awaiting order, sometimes you will arrive before they have completed preparing your order, so you will still have to wait. By issuing SETOBJACC only once into a dedicated pool, at any time before processing the data, is like a guarantee that the food order is ready whether you arrive immediately or sometime later; this would be like for every time you call ahead with your food order, they have completed the order before you /arrive/ to collect the take-out... even if you called-in your order from your cell phone while standing at the counter in front of the cashier.

Regards, Chuck