>>Hi Naomi,
>>
>>thanks, but most of those concentrate on geting the "max" record out of a given set. Nice insight anyway.
>>
>>My problem is that I somehow need to split rather then concentrate.
>>
>>The given set is
>>

>>P_IX,iMat,nQty
>>4,2,0.2
>>5,2,20
>>6,2,8
>>7,2,22
>>

>>
>>I need to take qty of 25
>>
>>so return should be
>>

>>P_IX,iBase,iMat,nQty
>>1,7,2,22
>>2,5,2,3
>>

>>or even better
>>

>>P_IX,iBase,iMat,nQty
>>1,7,2,22
>>2,6,2,3
>>

>>
>>because base record P_Ix=6 fits better
>>
>>so I need to slice the qty to be taken into pieces fitting to the amount in the source.
>>
>>The target take 22 out of record P_Ix=4 of the source and the rest of 3 out of the next.
>>
>>Agnes
>
>I see - a bit tricky. I solved a similar problem with recursive cte (see http://blogs.lessthandot.com/index.php/DataMgmt/DataDesign/cte-and-hierarchical-queries - though it doesn't work for best fitting).
>
>I think you may be better with dbase solution here rather than SQL.

Argh. This is beyond my level - and I can't do it with VFP anyway. So we stuck to SCAN and LOCATE. (-.-)

Agnes