Question

I’m new to Ninject so what I’m trying may not even be possible but I wanted to ask. I free-handed the below so there may be typos. Let’s say I have an interface:

public interface IPerson
{
    string FirstName { get; set; }
    string LastName { get; set;}

    string GetFullName();
}

And a concrete:

public class Person : IPerson
{
    public string FirstName { get; set; }
    public string LastName { get; set; }

    public string GetFullName()
    {
        return String.Concat(FirstName, " ", LastName);
    }
}

What I’m used to doing is something like this when I’m retrieving data from arrays or xml:

public IEnumerable<IPerson> GetPeople(string xml)
{
    XElement persons = XElement.Parse(xml);

    IEnumerable<IPerson> people = (
        from person in persons.Descendants("person")
        select new Person
        {
            FirstName = person.Attribute("FName").Value,
            LastName = person.Attribute("LName").Value
        }).ToList();

    return people;
}

I don’t want to tightly couple the concrete to the interface in this manner. I haven’t been able to find any information in regards to using Ninject with LINQ to Objects or with object initializers. I may be looking in the wrong places, but I’ve been searching for a day now with no luck at all.

I was contemplating putting the kernel into an singleton instance and seeing if that would work, but I’m not sure that it will plus I’ve heard that passing your kernel around is a bad thing. I’m trying to implement this in a class library currently. If this is not possible, does anyone have any examples or suggestions as to what the best practice is in this case? Thanks in advance for the help.

EDIT:
Based on some of the answers I feel I should clarify. Yes, the example above appears short lived but it was simply an example of one piece that I was trying to do. Let’s give a bigger picture. Say instead of XML I am gathering all my data through a 3rd party web service and I’m creating an interface for it, the data could be a defined object in the wsdl or it could sometimes be an xml string. IPerson could be used for both the Person object and a User object. I will be doing this inside of a separate class library, because it needs to be portable and will be used in other projects, and handing it to an MVC3 Web Application and the objects will be used in javascript as well. I appreciate all the input so far.

Answer 1

Your Person class is a short lived object and it doesn’t lend itself well to be used for dependency injection. Besides this, it doesn’t contain any behavior and is simply a POCO (plain old CLR object). Because POCOs don’t depend on anything worth abstracting, there is usually no reason to abstract them. In other words: With the given example. you don’t need an IPerson interface. You can work directly with the Person class throughout your application.

The GetPeople method though, could typically be part of a service that you would abstract away using your DI configuration. A service interface that contains the GetPeople(string xml) method however, would probably be a wrong abstraction, because this means you would always supply an xml string. When you have this XML string, would there be any reason to ever parse that XML string in any other way? It would be more convenient to have an IPersonRepository interface with a GetAllPeople() method. A given implementation could be the XmlPersonRepository that uses an XML data source to fetch the people (from disk, database or who knows what).