>>I am updating a series of PRG's to optimize them for speed and to get them ready to convert into classes.
>>
>>I am finding statement's like:
>>
>>
>>   replace all cpt_code with h->new_cpt for;
>>      h->new_cpt<>space(10).and..not.subs(h->new_cpt,1,1)$'QHZ'
>>
>>
>>My updated version is:
>>
>>
>>   REPLACE ALL Cpt_Code WITH Cdm.new_cpt;
>>     FOR NOT EMPTY(Cdm.new_cpt) AND NOT LEFT(Cdm.new_cpt,1) $ "QHZ"
>>
>>
>>
>>In the first example, you can see where I replaced the '<>space(10)' with 'NOT EMPTY()' and
the 'SUBSTR()' with 'LEFT()'
>>
>>
>>So my question is, in a program with many of these REPLACE statements, what kind of speed advantages will I see with these changes?
>
>You may actually see a performance hit. Using "<> space(10)" is optimizable where the Empty() function is not. Also the $ operator is not optimzable. So you have made a partially optimable (because 1 side of the AND is and 1 side is not) FOR clause not optimzable at all.
>
>FWIW, to make the statement fully optimzable, go back to the space(10) option and change the $ code to:

>NOT (Cdm.new_cpt = "Q" OR Cdm.new_cpt = "H" OR Cdm.new_cpt = "Z")

>If you choices are limited to only three, this may help with speed. Otherwise, partially optimzable is better than none.
>
>HTH.
You can also use Inlist() function, which is Rushmore optimizable. Note, also, that this syntax would work only if SET EXACT OFF and SET ANSI OFF (partial string compare).