Question

I have need to select a number of ‘master’ rows from a table, also returning for each result a number of detail rows from another table. What is a good way of achieving this without multiple queries (one for the master rows and one per result to get the detail rows).

For example, with a database structure like below:

MasterTable:     - MasterId BIGINT     - Name NVARCHAR(100)  DetailTable:     - DetailId BIGINT     - MasterId BIGINT     - Amount MONEY 

How would I most efficiently populate the data object below?

IList<MasterDetail> data;  public class Master {     private readonly List<Detail> _details = new List<Detail>();      public long MasterId     {         get; set;     }      public string Name     {         get; set;     }      public IList<Detail> Details     {         get         {             return _details;         }     } }  public class Detail {     public long DetailId     {         get; set;     }      public decimal Amount     {         get; set;     } } 

Answer 1

Normally, I’d go for the two grids approach – however, you might also want to look at FOR XML – it is fairly easy (in SQL Server 2005 and above) to shape the parent/child data as xml, and load it from there.

SELECT parent.*,        (SELECT * FROM child        WHERE child.parentid = parent.id FOR XML PATH('child'), TYPE) FROM parent FOR XML PATH('parent') 

Also – LINQ-to-SQL supports this type of model, but you need to tell it which data you want ahead of time. Via DataLoadOptions.LoadWith:

// sample from MSDN Northwnd db = new Northwnd(@'c:\northwnd.mdf'); DataLoadOptions dlo = new DataLoadOptions(); dlo.LoadWith<Customer>(c => c.Orders); db.LoadOptions = dlo;  var londonCustomers =     from cust in db.Customers     where cust.City == 'London'     select cust;  foreach (var custObj in londonCustomers) {     Console.WriteLine(custObj.CustomerID); } 

If you don’t use LoadWith, you will get n+1 queries – one master, and one child list per master row.