Fabio
1) Tested with your final solution, i runs in 4 seconds ( table has 250000 records )...I think the fact i had index on CDate helps.
2) In your sample code, if i have at say 9:00 Fabio on Monday and Jaime on Wednesday, they figure in the same row in the result, which is the desired behavior. In real application, this happens only 1% of the cases.
On the other hand, i don't care, since i now have a cursor with the data i want, i can just SCAN...ENDSCAN and put in the same row whenever applicable.

Thanks for this incredible lesson Fabio, it's pure art what you did :-)

Jaime


>>>>Fabio
>>>>Why is it you need it? I have an OperationID field which is the primary key for the table, but i don't think it's relevant in this case.
>>>>Here, the user will enter the date for Sunday ( let's call it SunDate ) and i just extract all records where BETWEEN( CDate, SunDate, SunDate+3 ) ordered by CTime.
>>>>Jaime
>>>
>>>

>>>SELECT ;
>>>	CTime ;
>>>,	MAX(IIF(DateW=1,Patient,SPACE(30)))	AS Sunday     ;
>>>,	MAX(IIF(DateW=2,Patient,SPACE(30)))	AS Monday     ;
>>>,	MAX(IIF(DateW=3,Patient,SPACE(30)))	AS Tuesday    ;
>>>,	MAX(IIF(DateW=4,Patient,SPACE(30)))	AS Wednesday  ;
>>>,	MAX(IIF(DateW=5,Patient,SPACE(30)))	AS Thursday	;
>>>,	MAX(IIF(DateW=6,Patient,SPACE(30)))	AS Friday	;
>>>,	MAX(IIF(DateW=7,Patient,SPACE(30)))	AS Saturday	;
>>>	FROM (SELECT	MAX(T.CTime) 	   CTime   ;
>>>	,		MAX(DOW(T.CDate))  DateW   ;
>>>	,		MAX(T.Patient)	   Patient ;
>>>	,		COUNT(*)           rowInGroup;
>>>			FROM theTable T ;
>>>			LEFT JOIN theTable RN ;
>>>				ON RN.OperationID <= T.OperationID ;
>>>				AND RN.CTime=T.CTime ;
>>>				AND DOW(RN.CDate)=DOW(T.CDate);
>>>			GROUP BY T.OperationID) GG;
>>>	GROUP BY CTime,rowInGroup
>>>

>>>if it is slow, a variation exists
>>
>>Hey Fabio
>>1) It's extremely slow ( my lamborgini computer is stucked on it already for 5 minutes and still didn't finish :-) ) even if i deleted the code for thursday friday and saturday
>
>These are too much complex things for VFP,
>only an advanced engine (at least SQL Server)
>it is able to use the indexes to optimize this query ( or that in the other message),
>but queries exist where also SQL Server surrenders him
>and use a slow filter on a products Cartesian.
>
>
>>2) i don't understand this join

FROM theTable T ;
>>			LEFT JOIN theTable RN ;
>>				ON RN.OperationID <= T.OperationID ;
>>				AND RN.CTime=T.CTime ;
>>				AND DOW(RN.CDate)=DOW(T.CDate)

>
>This build a row sequential number for avery T.CTime,DOW(T.CDate) couple,
>then : T.CTime,DOW(T.CDate),rowInGroup become a candidate key.
>At this point, GROUP BY CTime,rowInGroup
>it merge the lines with the same CTime,rowInGroup,
>and the operation of transposition of DOW(T.CDate)
>put every candidate item into a specific cell of the result.
>
>>Thanks
>>Jaime