tags:

views:

447

answers:

3

Hello, I'm having trouble with a query written in Linq and Lambda. So far, I'm getting a lot of errors here's my code:

int id = 1;
var query = database.Posts.Join(database.Post_Metas,
                                post => database.Posts.Where(x => x.ID == id),
                                meta => database.Post_Metas.Where(x => x.Post_ID == id),
                                (post, meta) => new { Post = post, Meta = meta });

I'm new to using Linq, so I'm not sure if this query is correct.

+8  A: 

I find that using the linq query syntax is much clearer and makes it easier to spot errors:

var id = 1;
var query =
   from post in database.Posts
   join meta in database.Post_Metas on post.ID equals meta.Post_ID
   where post.ID == id
   select new { Post = post, Meta = meta };

If you're really stuck on using lambdas though, your syntax is quite a bit off:

var id = 1;
var query = database.Posts    // your starting point
   .Join(database.Post_Metas, // the source table of the inner join
      post => post.ID,        // Select the primary key (the first part of the "on" clause in an sql "join" statement)
      meta => meta.Post_ID,   // Select the foreign key (the second part of the "on" clause)
      (post, meta) => new { Post = post, Meta = meta }) // selection
   .Where(postAndMeta => postAndMeta.Post.ID == id);    // where statement
Daniel Schaffer
Your post contains an easy to spot error. ;)
Mark Byers
Oh irony... I'm not seeing it... but that doesn't mean anything
Daniel Schaffer
@Daniel: The table is called `Post_Metas`, not `Meta`. The principle is correct though.
Mark Byers
You've answered my question! Thank you.
David
Oh oops! For some reason I'd thought that was the name of the join variable
Daniel Schaffer
@David, glad to help!
Daniel Schaffer
+2  A: 

Your key selectors are incorrect. They should take an object of the type of the table in question and return the key to use in the join. I think you mean this:

var query = database.Posts.Join(database.Post_Metas,
                                post => post.ID,
                                meta => meta.Post_ID,
                                (post, meta) => new { Post = post, Meta = meta });

You can apply the where clause afterwards, not as part of the key selector.

Mark Byers
+2  A: 

You could go two ways with this. Using LINQPad (invaluable if you're new to LINQ) and a dummy database, I built the following queries:

Posts.Join(
    Post_metas,
    post => post.Post_id,
    meta => meta.Post_id,
    (post, meta) => new { Post = post, Meta = meta }
)

or

from p in Posts
join pm in Post_metas on p.Post_id equals pm.Post_id
select new { Post = p, Meta = pm }

In this particular case, I think the LINQ syntax is cleaner (I change between the two depending upon which is easiest to read).

The thing I'd like to point out though is that if you have appropriate foreign keys in your database, (between post and post_meta) then you probably don't need an explicit join unless you're trying to load a large number of records. Your example seems to indicate that you are trying to load a single post and it's meta data. Assuming that there are many post_meta records for each post, then you could do the following:

var post = Posts.Single(p => p.ID == 1);
var metas = post.Post_metas.ToList();

If you want to avoid the n+1 problem, then you can explicitly tell LINQ to SQL to load all of the related items in one go (although this may be an advanced topic for when you're more familiar with L2S). The example below says "when you load a Post, also load all of its records associated with it via the foreign key represented by the 'Post_metas' property":

var dataLoadOptions = new DataLoadOptions();
dataLoadOptions.LoadWith<Post>(p => p.Post_metas);

var dataContext = new MyDataContext();
dataContext.LoadOptions = dataLoadOptions;

var post = Posts.Single(p => p.ID == 1); // Post_metas loaded automagically

It is possible to make many LoadWith calls on a single set of DataLoadOptions for the same type, or many different types. If you do this lots though, you might just want to consider caching.

Damian Powell