Cetin,
>> We have this program in our library.
>>
>>*!******************************************************************************
>>FUNCTION aparser
>> * Description.......: Function parses a text string into an array.
>> * : Each element of the array contains one word...
>> * Calling Samples...: dimension laWords[1]
>> * : lcWordString = "Some text string"
>> * : lnWords = aparser(@laWords, lcWordString)
>> * Parameter List....: paWords, pcWordString, pcDelimiter
>> ********************************************************************
>> lparameter paWords, pcWordString, pcDelimiter
>> local lnWords, lni
>> external array paWords
>>
>> if not "FOXTOOLS.FLL" $ set("library")
>> set library to foxtools.fll additive
>> endif
>>
>> pcDelimiter = iif(empty(pcDelimiter), space(1), pcDelimiter)
>>
>> lnWords = Words(pcWordString, pcDelimiter)
>>
>> IF lnWords>0 && Not empty string
>> dimension paWords[lnWords]
>>
>> for lni = 1 to lnWords
>> paWords[lni] = WordNum(pcWordString, lni, pcDelimiter) && Array of words
>> endfor
>> ELSE
>> dimension paWords[1] && For empty string lnWords=0
>> paWords[1]=""
>> ENDIF
>> return lnWords
>>
>>In you case Delimiter will be ','.
>>
>>HTH
>
>

---

FYI: Doing parsing with lowlevel instead of foxtools word* functions rewards you with speed :) Even another parser fll is faster because it uses the logic that it doesn't have to rewind each time it has to find next word as foxtools does.
>Tested speed is about 1Mb/sec (parse, sort, get into array + into a cursor - K6-2-333 64Mb).

If I understand you correctly, you think that our approach is slower than others. I totally agree. Actually, this is not the program I created, we already had this in our program libraries, I just use it.

For my regret, I didn't see the whole thread and I'm not a PUTM, so I don't know what you and others suggested. If you know better method just give me the code, please.