select top n results in a group; count(*)<=n, count smallest largest etc. problem

Two tables: CompanyPrices(CompanyID, ProductID, Price), CompanyRegion(CompanyID, Region)
ProductID is the primary key.

I want to get 10 smallest prices in each Region. In other words, I am looking for 10 cheapest prices in each region. So, if there are 20 regions, I should get excatly 200 rows having prices for products from 200 companies if there were at least 10 companies in each region.

I tried the follwoing, but get incorrect results.

select S1.Region, S1.Price from (select CompanyPrices.*, Region from CompanyPrices inner join CompanyRegion on CompanyPrices.CompanyID=CompanyRegion.CompanyID) S1 inner join
(select CompanyPrices.*, Region from CompanyPrices inner join
CompanyRegion on CompanyPrices.CompanyID=CompanyRegion.CompanyID) S2 on S1.Region = S2.Region
group by S1.Region, S1.Price having count(*)<=10 order by S1.Region, S1.Price

However, if I want to get 10 cheapest products for each company, the above sql works by modifying the join condition. Instead of S1.Region = S2.Region , I use S1.CompanyID = S2.CompanyID and I get correct results for 10 cheapest products for each company.

select S1.CompanyID, S1.Price from (select CompanyPrices.*, Region from CompanyPrices inner join CompanyRegion on CompanyPrices.CompanyID=CompanyRegion.CompanyID) S1 inner join
(select CompanyPrices.*, Region from CompanyPrices inner join
CompanyRegion on CompanyPrices.CompanyID=CompanyRegion.CompanyID) S2 on S1.CompanyID = S2.CompanyID
group by S1.CompanyID, S1.Price having count(*)<=10 order by S1.CompanyID, S1.Price

Of course, the join in S1 and S2 with the CompanyRegion table is not necessary in the second query.

I am not sure what is wrong in the first query and why it does not work when the second one works. Could someone please help in making the first query work to give me correct results?

Thanks Received on Sun Jan 19 2003 - 16:42:27 CET