再谈谈LINQ TO DataSet
首先，这个功能是通过扩展方法来实现的，需要引用System.Data.DataSetExtensions这个程序集
其次，针对DataSet有两种情况
第一种是非类型话的DataSet，下面代码是一个示例
DataSet ds = new DataSet();
using (SqlConnection conn = new SqlConnection("server=(local);database=northwind"))
{
using (SqlCommand cmd = conn.CreateCommand())
{
cmd.CommandText = "SELECT * FROM Customers";
SqlDataAdapter adapter = new SqlDataAdapter(cmd);
adapter.Fill(ds, "Customers");
cmd.CommandText = "SELECT * FROM Orders";
adapter.Fill(ds, "Orders");
}
}
DataRow defaultRow=ds.Tables["Orders"].NewRow();
defaultRow["OrderID"] = -1;
defaultRow["OrderDate"] = DateTime.Now;
var query=from c in ds.Tables["Customers"].AsEnumerable()
join o in ds.Tables["Orders"].AsEnumerable() on c.Field<string>("CustomerID") equals o.Field<string>("CustomerID") into g
from item in g.DefaultIfEmpty(defaultRow)//这里是为了实现左外连接，因为有的客户可能没有订单
select new{
CustomerID=c.Field<string>("CustomerID"),
CompanyName=c.Field<string>("CompanyName"),
OrderID=item.Field<int>("OrderID"),
OrderDate=item.Field<Nullable<DateTime>>("OrderDate")
};
foreach (var item in query)
{
Console.WriteLine(item);
}
现在我们可以针对不同的表直接进行JOIN或者其他所有的LINQ操作，例如ORDER BY，GROUP BY等等
第二种是针对类型化的DataSet，也就是我们常说的强类型数据集。使用它们太方便了，就是对象操作。
所以，强烈建议要创建强类型的数据集
如果是使用强类型的数据集，则读取数据及处理数据的工作都会变得很方便和可靠
NorthwindDataSet northwind = new NorthwindDataSet();
NorthwindDataSetTableAdapters.CustomersTableAdapter a = new NorthwindDataSetTableAdapters.CustomersTableAdapter();
a.Fill(northwind.Customers);
NorthwindDataSetTableAdapters.OrdersTableAdapter o = new NorthwindDataSetTableAdapters.OrdersTableAdapter();
o.Fill(northwind.Orders);
NorthwindDataSet.OrdersRow defaultrow=northwind.Orders.NewOrdersRow();
defaultrow.OrderID=-1;
defaultrow.OrderDate=DateTime.Now;
var query = from customer in northwind.Customers
join order in northwind.Orders on customer.CustomerID equals order.CustomerID into orders
from item in orders.DefaultIfEmpty(defaultrow)
select new
{
CustomerID = customer.CustomerID,
CompanyName = customer.CompanyName,
OrderID = item.OrderID,
OrderDate=item.Field<Nullable<DateTime>>("OrderDate")
};
