>>>>I'm still using VFP8 here, take this into account. Perhaps, in VFP9 filtered indexes are Rushmore optimizable.
>>>
>>>On this ( SEEK + SCAN WHILE ) code , every optimization is out of the game.
>>>
>>>Nadya,
>>>
>>>INDEX conditions1 FOR conditon2
>>>
>>>with
>>>SET ORDER
>>>SEEK condition1
>>>SCAN WHILE condition1
>>>
>>>
>>>is the faster possible code into VFP with:
>>>- with TALK OFF
>>>- table/index cached.
>>>
>>>When you do intensive read/updated, cache operations can change timing results randomly.
>>>
>>>Try to decrease/increase the SYS(3050,1) value.
>>
>>Thanks. I was so disappointed of the performance, that I already switched to SCAN FOR without order set.
>
>Hi Nadya,
>tell me the results.
>
>But, my impression is that you want to obtain what VFP cannot gives to you.
>Shooting to a rule + a trigger for every record,
>I doubt that you can arrive more than 10000 record/s with direct code.
My yesterday's cases showed ~100s. per 50000 recs (I did replace field with field), while native RI code gave ~ 50 sec. per 50000 recs. I haven't re-tested today yet, yesterday I did one more obvious optimization and fixed a typo in one place. I'll give you my results later on.
If it's not broken, fix it until it is.
My Blog