Can I use my own Model classes with Linq to SQL?

Question

Hi

I have a fairly large application, where my data access strategy has always been quite old school:

I have 4 Stored Procedures per Table: TableName_Select, TableName_Insert, TableName_Update, TableName_Delete.

I have a MsSql#DOMAIN#Service class for each logical domain of my application. In that class I have corresponding method to Select, Insert, Update and Delete something in the database.

I want to maintain my own Model classes in a seperate class library project, along with service contracts for WCF services. That enables me to reference only the naked Models, Service Contracts and WCF Service Clients from e.g. a web app or a windows app. (I don't use the auto-generated WCF Service Clients either)

Can I use my own Model classes along with Linq-to-SQL - or do I have to use the autogenerated Models from Linq-to-SQL and then map them over to my own Models in my data access layer before returning them back through the WCF service?

Answer 1

You don't have to use the auto-generated types with a Linq To Sql data context, but by default the types that you use need to have all the attribute annotations that you'll see on them.

However, Linq to Sql also supports custom mapping (the default being to use the Attributes), where types are mapped to tables and their properties mapped to columns using a MappingSource which is provided to the DataContext on construction.

L2S comes with two - the AttributeMappingSource (used implicitly) and the XMLMappingSource. It's possible with the second of these that you can hand-crank a Linq To Sql DataContext that exposes EntitySets of your own types.

You could even write your own MappingSource - but I think the XML one probably covers most needs.

Answer 2

We have done something similar in our applications. We wrote our own code generator to generate our L2S classes and also what we call our "application" entities. These are much more lightweight than the L2S classes. They are used at the application level to pass data back and forth to our back-end. Each L2S entity class has a built in application entity equivalent where automatic mapping occurs. What I mean is that whenever a L2S entity has data stored to its properties, the values are automatically copied to the respective application entity property. We then have a method on each L2S entity that allows us to retrieve the associated application entity.

So, the short answer is yes, you can use your own classes in conjunction with the L2S entities.