>Cursor is fastest in my quick benchmark (by 3 seconds on 100,000 iterations), but I have to loop at least 10,000 times to see any significant difference between the 2 methods. For your 1,000 records, any way would take about the same time.
CLEAR
>
>LOCAL ARRAY laTest [1000]
>
>CREATE CURSOR Test (Test I)
>
>FOR I = 1 TO 1000
> laTest[I] = I
> INSERT INTO Test VALUES (I)
>ENDFOR
>
>? "Array"
>? DATETIME()
>
>FOR J = 1 TO 100000
> FOR I = 1 TO 1000
> I = laTest[I]
> ENDFOR
>ENDFOR
>
>? DATETIME()
>
>? "Cursor"
>? DATETIME()
>
>FOR J = 1 TO 100000
> SCAN
> I = Test.Test
> ENDSCAN
>ENDFOR
>
>? DATETIME()
>If I had 1000 values that I need to loop through (visit every node sequentially, not skip around) that needed very fast processing and they do not need disk storage... which would be faster... an array or a cursor?
With exact same code, before any change, array version was faster 1 second. After adding m. prefix, array was faster several seconds.
Cetin
