I just tried it. To replace value with 50000 records (the same value to the same value) it originally took 91 sec. and now it's 35 sec., with native RI is taking 51 sec. That's definitely a huge improvement.

Let me check various conditions and if nothing fails I owe you big < g >

>Hi Nadya,
>
>I am getting more doubtful on the validity of my previous assumption:
>Is the "Currentrecord" in RIDefinitions guaranteed to stay even when triggers maybe call RI recursively inside each function ?
>Since RIDefinitions is always used as alias and the recordponter insn't saved/reset,
>any recursion could be dangerous to the correct coverage of the scan-loops.
>
>Check the out the following (mostly done as finger flexing while waiting for another nachine to finish processing -
>neither checked nor guaranteed - use to get ideas or at you own risk <g>), especially check the parts with cL_USE_AGAIN...
>
>Hopefully worth more than you paid for it ;-)
>
>regards
>
>thomas
>