Suhas
If you would explain what you're trying to do we can show you a better way to do it. I can't see why you need 2 SQL and a double nested scan with an internal seek. That has got to be far too complex.
Show us a small sample of the data.
symbol, open, date
then show us the expected array.
What is bhav_data.pk again?
By the way, what does the fll do? It may be possible to get the entire result in a single SQL, bypassing the fll.
>I use the array to feed to a fll.
>
>So i decided to use the CDX and use seek
>
>That Is really as fast as building a IDX and running.
>Avoided the use of SQL alto gether in the scan..endscan
>
>
>COde
>
>
>Select dist symbol from bhav_data into cursor dist1 nofilter
>select date from bhav_data group by 1 order by 1 into cursor date_temp nofilter
>select Dist1
>scan
> dimension aopen[1]
> arraycount = 0
> select date_temp
> scan
> if seek(dtos(date)+dist1.symbol,'bhav_data','pk')
> arraycount= arraycount+1
> dimension aopen[arraycount]
> aopen[arraycount] = bhav_data.open
> endif
> endscan
>
>* feed the array adate
>
>endscan
>
>That is fast , but not really fast as rebuilding idx in code but I feel with a marginial difference in time (about .5x difference) and well faster that using SQL abiout 3x
>
>
>
>Thanx all for putting me use the CDX
>
>Suhashegde