joining in Linq

Question

When i declare

 int[] a = { 1, 2 ,10,30};

 int[] b = { 10, 20, 30 };

  var q = from item1 in a
                     join item2 in b
                     on item1 equals item2
                     into g
                     select g

1)What is actually getting selected into g ? It is difficult to understand the into keyword. if you give example to explain the keyword "into",i will be pleased.

2)

 How do the following code actually projected ?

1    from a in db.coll1
2    join b in db.coll2 on a.PK equals b.PK into b_join
3    from b in b_join.DefaultIfEmpty()
4    where
5      b.PK == null 
6   select new {
7      a.PK,
8      a.Value,
9      Column1 = (System.Int32?)b.PK,
10      Column2 = b.Value
    }

in line 2 we are using "b' to select the item in line 3 also we are using the same b ,does it mean we are overriding the data we selected at line 2 ?

Answer 1

If you use .NET Reflector on your compiled code, you can see that your first query is equivalent to the following:

IEnumerable<IEnumerable<int>> q = a.GroupJoin(b,
   item1 => item1,
   item2 => item2,
   (item1, group) => group);

Note that you are performing a group join, but only returning the groups. Compare this with an ordinary join:

IEnumerable<int> q = a.Join(b,
    item1 => item1,
    item2 => item2,
    (item1, item2) => item2);

This returns the elements in b that matches to each a, if any such element exists.

Answer 2

1) You are doing a group join. g will be an IGrouping for each element in a. g.Key will be the element from a, and g (which implements IEnumerable) will contain all of the matching elements from b. To understand this fully, try it out in LINQPad. Also, read the documentation.

2) The into keyword merges the first b into b_join, which will be an IGrouping as I described above. After that, the first b is no longer in scope.