Memory when using locate() over big tables

[Rulikkk]

Hello.
I have a search form in my application in a grid, that is using TOracleDataSet's locate() method.
When the table is huge, my application takes up to 2gb of memory.
Is there any way to cure that?
I've tried using unidirectional, but in that case, after the search I can't scroll my grid up, it says "The record ... is not available in undirectional mode"
Is there any way to make data set free records that it has checked during search, and then to reload them from server, in case they are needed?

 

[Marco Kalter]

For large result sets, use a query with an appropriate where clause instead. This is a much better solution than a local search over the complete result set.

 

[Rulikkk]

Originally posted by Marco Kalter:
For large result sets, use a query with an appropriate where clause instead. This is a much better solution than a local search over the complete result set.

I know. But I need a solution for this problem, and not to avoid it.

 

[Marco Kalter]

I'm afraid the only solution is to avoid it though.

 

[Rulikkk]

Ok, thank you for your answers.