实现LINQ提供程序

Question

I'm currently implementing a LINQ provider for my own educational purposes. I have managed to get Count() extension to work recently, so far so good.

Now my question is not a cry for help, but just a request for some clarification.

There are two interfaces to be implemented in order to create the provider: IQueryProvider and something like IOrderedQueryable<> . MSDN makes clear how one implements them, but one point is still confusing me.

Why these interfaces are implemented by separate classes , even though each IOrderedQueryable instance refers own IQueryProvider instance and both objects actually (indirectly) refer the same data?

Do they really need to be separated?

Furthermore. I am able to combine them like this: class Source<RowContract> : IQueryProvider, IOrderedQueryable<RowContract> - in order to simplify type information access. This implementation works properly now and looks more simple and clear than "separate-classes" approach.

I am wondering if there is a flaw in my combined implementation. Or, maybe it's valid?

Any explanation would be appreciated greatly.

Answer 1

As mentioned on msdn IQueryProvider is focused on creating and executing the query. Whereas IQueryable is the thing being queried. Rolling it all together may put similar code together, but it ultimately doesn't respect separation of concerns.