I suppose that may simplify the code a little. How about the randomness of the numbers that are returned, do you believe this to be as random as it can be? The help file indicates calling RAND 1st with a -1 to achieve the most random result.
Jeff
>If you just want to get ca. 5% of the population, you can make it a little simpler. There is no need to multiply rand() with any constant; just compare if rand() < 0.05 for all employees:
>
>
>rand(-1)
>scan
> if rand() < 0.05
> * process this employee
> endif
>endscan
>
>
>HTH,
>
>Hilmar.
>
>
>>Hello all,
>>
>>I have a program that selects employees for random drug testing. I would like to verify that my code produces the most random number possible. I just pasted the code below, any comments are appreciated.
>>
>>
>>Thanks, Jeff
>>
>>_________________________________________________________________________
>>
>>lparameters lcPlantNo
>>
>>* get only active employee #'s into the temporary
>>* table
>>select prempl,first,last ;
>> from prempl01 where left(prempl,1) = lcPlantNo ;
>> and status = "A" ;
>>union all ;
>>select prempl,first,last ;
>> from prempl02 where left(prempl,1) = lcPlantNo ;
>> and status = "A" ;
>> into cursor tmpEmpl
>>
>>select tmpEmpl
>>
>>nRecs = reccount()
>>* the number of tests to perform is 5% of
>>* the total population for this plant
>>nNumTests = round(nRecs * .05,0)
>>nTested=0
>>* initialize the seed value
>>nEmplNo = rand(-1)
>>do while nTested < nNumTests
>>
>> * get the random employee number
>> nEmplNo = int(nRecs * rand() + 1)
>> * random sort for printout
>> nRandSort = rand()*100
>>
>> if nEmplNo < 1 or nEmplNo > nRecs
>> loop
>> endif
>>
>> * save the employee to the temp file
>> goto nEmplNo in tmpEmpl
>> select testnumbers
>> seek tmpEmpl.prempl
>> if !found()
>> * employee not already included in this test
>> nTested = nTested + 1
>> append blank
>> replace empid with tmpEmpl.prempl
>> replace last with tmpEmpl.last
>> replace first with tmpEmpl.first
>> replace randsort with nRandSort
>> endif
>>enddo
>>
>>use in tmpEmpl
>
