Retrieve data from stored procedure which has multiple result sets

Given a stored procedure in SQL Server which has multiple select statements, is there a way to work with those results separately while calling the procedure?

For example:

alter procedure dbo.GetSomething
as
begin
    select * from dbo.Person;
    select * from dbo.Car;
end;

In .NET, if I call this proc, I can use a SqlDataReader to move between the two result sets, so I can easily retrieve all people and cars. In SQL however, when I execute the proc directly, I get both result sets.

If I call:

insert @myTempTable
    exec dbo.GetSomething;

Then it errors because the column definition doesn't match. If by some chance Person and Car have the same columns, it concatenates the two together, and @myTempTable gets all records from both tables, which obviously is no good either.

I can define new custom types representing the two result sets, and make those output parameters instead of having the multiple select statements, but I'm wondering if there's a better way - some way of pulling both results into temporary tables, or looping through the results, or something.

EDIT

Actually, after looking more closely, even output table parameters won't solve this - they're readonly, and that's still true in SQL 2012. (Connect ticket asking for this to be added)

Solution 1:

String myConnString  = "User ID="username";password="password";Initial Catalog=pubs;Data Source=Server";
SqlConnection myConnection = new SqlConnection(myConnString);
SqlCommand myCommand = new SqlCommand();
SqlDataReader myReader ;

myCommand.CommandType = CommandType.StoredProcedure;
myCommand.Connection = myConnection;
myCommand.CommandText = "MyProc";

try
{
    myConnection.Open();
    myReader = myCommand.ExecuteReader();

    while (myReader.Read())
    {
        //Write logic to process data for the first result.   
        }

    myReader.NextResult();
    while (myReader.Read())
    {
        //Write logic to process data for the second result.
    }
}

Solution 2:

In TSQL land, you're stuck.

Here is a trick (some may call semi-hacky) way that I used one time.

/*  START TSQL CODE */

/*  Stored Procedure Definition */

Use Northwind
GO


IF EXISTS 
    (
    SELECT * FROM INFORMATION_SCHEMA.ROUTINES
    WHERE ROUTINE_TYPE = N'PROCEDURE' and ROUTINE_SCHEMA = N'dbo' and ROUTINE_NAME = N'uspOrderDetailsByCustomerId'  
    )
BEGIN
    DROP PROCEDURE [dbo].[uspOrderDetailsByCustomerId]
END


GO

CREATE Procedure dbo.uspOrderDetailsByCustomerId
(
  @CustomerID nchar(5)
, @ResultSetIndicator smallint = 0
)
AS

BEGIN

    SET NOCOUNT ON



    /* ResultSet #1 */

    if (@ResultSetIndicator = 0 OR @ResultSetIndicator = 1)
    BEGIN 
        SELECT 
            c.CustomerID, c.CompanyName /*,c.ContactName,c.ContactTitle,c.[Address],c.City,c.Region,c.PostalCode,c.Country ,c.Phone,c.Fax */
        FROM 
            Customers c 
            JOIN Orders o ON c.CustomerID = o.CustomerID 
        WHERE 
            c.CustomerID = @CustomerID
    END


    /* */
    /* ResultSet #2 */ 

    if (@ResultSetIndicator = 0 OR @ResultSetIndicator = 2)
    BEGIN 

        SELECT o.OrderID,o.CustomerID /* ,o.EmployeeID,o.OrderDate,o.RequiredDate,o.ShippedDate,o.ShipVia ,o.Freight,o.ShipName,o.ShipAddress,o.OrderID,o.CustomerID,o.EmployeeID,o.OrderDate  */
        FROM 
            Orders o 
         WHERE 
            o.CustomerID = @CustomerID
        ORDER BY 
            o.CustomerID , o.OrderID 

    END


    /* */
    /* ResultSet #3 */

    if (@ResultSetIndicator = 0 OR @ResultSetIndicator = 3)
    BEGIN 
         SELECT od.OrderID,od.ProductID /* ,od.UnitPrice,od.Quantity,od.Discount  */
         FROM 
            [Order Details] od 
         WHERE 
            exists (select null from dbo.Orders  innerOrds where innerOrds.OrderID = od.OrderID and innerOrds.CustomerID = @CustomerID )
         ORDER BY 
            od.OrderID 

    END

    SET NOCOUNT OFF


END

GO 
/* Get everything */


exec dbo.uspOrderDetailsByCustomerId 'ALFKI'




    IF OBJECT_ID('tempdb..#TempCustomer') IS NOT NULL
    begin
            drop table #TempCustomer
    end


    CREATE TABLE #TempCustomer
    ( 
      [CustomerID] nchar(5)
    , [CompanyName] nvarchar(40)
    )

INSERT INTO #TempCustomer ( [CustomerID] , [CompanyName])
exec dbo.uspOrderDetailsByCustomerId 'ALFKI' , 1

Select * from #TempCustomer



    IF OBJECT_ID('tempdb..#TempOrders') IS NOT NULL
    begin
            drop table #TempOrders
    end


    CREATE TABLE #TempOrders
    ( 
        OrderID int
      , [CustomerID] nchar(5)

    )

INSERT INTO #TempOrders ( OrderID , [CustomerID] )
exec dbo.uspOrderDetailsByCustomerId 'ALFKI' , 2

Select * from #TempOrders






    IF OBJECT_ID('tempdb..#TempOrderDetails') IS NOT NULL
    begin
            drop table #TempOrderDetails
    end


    CREATE TABLE #TempOrderDetails
    ( 
        OrderID int
      , [ProductID] int

    )

INSERT INTO #TempOrderDetails ( OrderID , [ProductID] )
exec dbo.uspOrderDetailsByCustomerId 'ALFKI' , 3

Select * from #TempOrderDetails


    IF OBJECT_ID('tempdb..#TempOrderDetails') IS NOT NULL
    begin
            drop table #TempOrders
    end


    IF OBJECT_ID('tempdb..#TempOrders') IS NOT NULL
    begin
            drop table #TempOrders
    end



    IF OBJECT_ID('tempdb..#TempCustomer') IS NOT NULL
    begin
            drop table #TempCustomer
    end

Solution 3:

It seems like there's no good simple way to do this, without a hack or a major paradigm shift. It looks like the best way is to just split out the original procs and end up with one more proc than before:

Old way:

create procedure dbo.GetSomething
as
begin
    select * from dbo.Person;
    select * from dbo.Car;
end;

New way:

create procedure dbo.GetPeople
as
begin
    select * from dbo.Person;
end;

create procedure dbo.GetCars
as
begin
    select * from dbo.Car;
end;

-- This gives the same result as before
create procedure dbo.GetSomething
as
begin
    exec dbo.GetPeople;
    exec dbo.GetCars;
end;

Then when I'm in a different proc and need both result sets, I'd just have to call them one at a time.