For one to one relationships things are easy.
When it comes to one to many or many to many problems appear...
I am not using an ORM tool now for many reasons and i am wondering when i want to get data whether it is better to reassemble one to many relationship using multiple queries or in code..
For example.. Having a class Category and a class Product...
A product table has a collumn for category id (one category many products).
So for my full catalog is it better to execute 2 queries to get the categories and products i want and then populate for each category its products List ? (It is very easy with LINQ) ..
Or i should call query for each category ? Like select id from products where category_id=5;
Also i dont know how to name the functions like to set whether i want to fetch the other side of the relationship or not..
You should always use the least number of queries possible to retrieve your data. Executing one query per category to load the products is known as the N+1 problem, and can quickly cause a bottleneck in your code.
As far as what to name your methods that specify your fetch plans, name them after what the method actually does, such as IncludeProducts or WithProducts.
If you want to retrieve all categories and all their products, you can either select all categories and then select all products in two queries, or you can select in one query and group.
To use just one query, select an inner join for your two tables
SELECT c.*, p.*
FROM Category c INNER JOIN Product p ON c.CategoryId = p.CategoryId
and then construct business objects from the resulting dataset
result.GroupBy(r => r.CategoryId).Select(group =>
new Category(/* new Category using c.* columns */)
{
Products = /* new list of Products from p.* values */
});
But I have to ask - why aren't you using an ORM?
Related
I have:
Simple search engine which is a dynamically built sql query. From this query I'm receiving product Id and Points (search rank). This query is based on search configuration. User can chose which fields he want to search and etc.
IQueryable where are my products with filters applied and some additional data joined.
What I want to do:
Join results from Sql Query with IQueryable so I will have my products that fulfill search conditions and other filters and have Points joined. I need to get my results as IQuryable
I have tried to rewrite this search engine to linq but I ended up with really huge and slow queries.
I've tried to fire up my sql query and get it as IQueryable and then join it to my IQueryable. That didn't worked out got some error.
Any tips?
EDIT:
Sample SQL Search Query;
SELECT Id, SUM(Points) AS Points
FROM (
SELECT tw_Id Id ,500 AS Points
from tw__Towar
Where
tw_Nazwa Like '%WA20-800%'
union all
SELECT tw_Id Id ,2500 AS Points
from tw__Towar
Where
tw_Nazwa Like '%WA20-800%'
union all
SELECT tw_Id Id ,5000 AS Points
from tw__Towar
Where
tw_Symbol Like 'WA20-800 %') sub
group by Id
The code gets way more complicated when there are more then one keyword.
I've tried to execute it in way:
dbcontext.Database.SqlQuery<PointsId>(query).AsQueryable();
and then join it to my IQueryable. But it didn't worked.
I was practicing today when I realized that there are two ways linq to sql can retrieve data from db, I created two datagrid and used the two different ways to populate each of these datagrids and they produced the same result.
The first method is using joins to get data from related tables, and the other methods is using linq query like an object to access related tables. The code is shown below:
NorthWindDataContext dbContext = new NorthWindDataContext();
var orders = from ord in dbContext.Orders
select new { ord.ShipCountry , ord.Customer.ContactName};
var orders2 = from ord in dbContext.Orders
join cust in dbContext.Customers on ord.CustomerID equals cust.CustomerID
select new
{
ord.ShipCountry, cust.ContactName
};
var data = orders2;
DataGrid.ItemsSource= orders;
DataGrid2.ItemsSource = orders2;
My question like the title is if it is entirely necessary to use joins, because I find them really cumbersome to use sometimes.
You need to use something that gets you from the order to the customer.
Join can do this. This is how the second query works.
Having the order "know" about the customer can do this. This is how the first query works.
If your data provider is aware of the connection between order and customer then these will amount to the same thing.
If your data provider is not aware of the connection, then the approach in the first example would result in N + 1 look ups instead of 1.
A linq-friendly ORM will generally be aware of these connections as long as the appropriate relationship-marking attributes are present (just what that is differs between Linq2SQL, EF, NHibernate, etc.).
It's still important to know the join approach for cases where either the relationship isn't known about by the provider, or you have a reason to join on something other than a foreign-key relationship.
The answer is "sort of". Since you're using an ORM such as Linq-to-Sql, no you don't directly need to call join within your linq queries to accomplish what you're trying to do.
However, when the ORM activates the query it will generate actual SQL code that'll have a join statement in it to get the results you're querying. Since you're using an ORM though, the data returned is mapped to objects, and since Customer has a relationship between the objects, the relationship will also be translated to from the database INTO the objects.
ord.Customer.ContactName
The above statement is most likely translated to a JOIN statement performing an INNER JOIN between Customer & Orders.
Due to this, both of your LINQ queries most likely generating similar SQL queries. Both of which has a JOIN statement in them. Because the relationships between your objects also exists within the database (and everything is mapped together showing this relationship) you don't directly need to use join within a LINQ statement.
I am just wondering about how we can determine whether to use join or not in linq to sql.
Eg. let say if we have two tables like this
Table 1 Customer
id
name
Table 2 addresstype
id
address1
customerid
and
var address = from cu in Customer
from ad in addresstype
where cu.id == ad.customerid
select ad;
or
var address = from cu in Customer
join ad in addresstype on cu.id equals ad.customerid
select de;
Is both way are the same. Is there any difference in performance?
Also the second method, will it come up with an error if there isn’t any matching?
Are you using linq to entities or linq to SQL? If its the former then you can avoid both of these by defining your relationships in the model and using navigation properties. This would be the clearest way of doing things
Basically, these two LINQ queries are equivalent to the following SQL queries:
select ad.*
from Customer cu, AddressType ad
where cu.ID == ad.CustomerID -- I assume this was meant by the OP
and
select ad.*
from Customer cu
inner join AddressType ad on cu.id = ad.CustomerID;
The difference between these two queries is mostly semantic, since the database will do the same thing in both cases and return a same result set for both queries.
I would prefer the join syntax in both SQL and LINQ since it defines an explicit relationship between the two tables/entities, that is only implied in the join-less version.
These are seems same query, they return same result but I don't know which one can be a faster, it should be bench marked.
But, In the case of linq2sql I prefer correlated subquery over join, because currently if you want t check the equation two element you should use syntax of:
new {X,Y} equals new {X',Y'}
in join and if you have more than this equations you should convert it to nested query. So I Prefer to have a more readable code which uses minimum differences in difference actions.
To throw a third and more prefered method into the mix with LINQ to SQL, use associations between the tables (even if you don't have them set up in your database). With that in place, you can navigate the object graph rather than using joins:
var query = from cu in Customer
from ad in cu.Addresses
select ad;
Note: when querying the object graphs, LINQ to SQL translates the join into a left outer join where-as the join/where syntax by default is an inner join.
Joins in LINQ should be used when there isn't a natural relationship between the objects. For example, use a join if you want to see the the listing of stores that are in the same city as your customers. (Join Customer.Address.City with Store.Address.City).
There should not be a difference between these two queries. I actually wondered this question myself a few months ago. I verified this through LINQPad. It's a free tool that you can download and actually see the generated SQL of any LINQ query (this is the query that is sent to the database).
The generated SQL should be the same for these two queries.
If you're doing this through Visual Studio, there is also a way you can see the generated SQL as well.
I keep tables on different .sdf files because it's easy to manage them, ie; back up only changed db file, etc, plus in future db size might bigger and there is -4GB limit-
I need to join the tables and this will be my first -possibly LINQ- attempt. I know there are tons of examples/documents but a simple example would be nice to start.
This is the query for MS SQL Server:
SELECT personID, personPin, personName, seenTime
FROM db1.personList
LEFT JOIN db2.personAttendances on personID = seenPersonID
ORDER BY seenTime DESC
I think LINQ will be the way to go as you're querying across 2 different contexts. LINQ joins are quite easy: http://msdn.microsoft.com/en-gb/vcsharp/ee908647
Something like:
var q = from c in db1Context.personList
join p in db2Context.personAttendances on c.personID equals p.seenPersonID
select new { Category = c, p.ProductName };
I don't think SqlCE supports linking at the Db (SQL) level.
That means you'll have to use Linq-to-Objects. The example query has no WHERE clause so you can simply load the entire tables into Lists. But when the datasets get bigger that may not be acceptable.
I'm trying to create the equivalent of the below using NHibernate. I've done all the mappings using fluent and I can do the basic queries just fine but I have no idea how to do this.
-**Product Table**
Reference
Title
Subjects (Many to Many relationship)
Price
-**Subject table**
SubjectID
Name
-**SubjectToProductMapping Table**
Reference
SubjectID
Now I need to do this:
SELECT *
FROM Product
WHERE Reference IN
(Select Reference FROM SubjectToProductMapping WHERE SubjectID = #SubjectID)
Baring in mind the Product table has been simplified a great deal for the post and that I would prefer to use an IN statement to keep the rest of the query simpler. I would ideally like to create the query using Criteria becuase I will be using Criteria to page the results.
Thanks in advance
Why would you use an in when a join would suffice? provided your Products class has a mapped collection of subjects then you could just use this Criteria
IList<Product> results = session.CreateCriteria(typeof(Product))
.CreateCriteria("Subjects", JoinType.Join)
.Add(Resitctions.Eq(Projections.ID, subjectID))
.List<Product>();