Web API ideas

webapi_experiment1.cs

/*
<?xml version='1.0' encoding='UTF-8'?>
<feed xmlns="http://www.w3.org/2005/Atom"
      xmlns:as="http://atomserver.org/namespaces/1.0/"
      xmlns:os="http://a9.com/-/spec/opensearchrss/1.1/">
  <os:totalResults>65801</os:totalResults>
  <os:startIndex>0</os:startIndex>
  <os:itemsPerPage>2</os:itemsPerPage>
  <as:endIndex>153</as:endIndex>
  <link href="/myserver/v1/widgets/acme?start-index=153&amp;max-results=2" rel="next" />
  <link href="/myserver/v1/widgets/acme?start-index=0&amp;max-results=2" rel="self" />
  <author>
    <name>AtomServer APP Service</name>
  </author>
  <title type="text">acme entries</title>
  <updated>2007-10-05T19:17:42.750Z</updated>
  <id>tag:atomserver.org,2007:v1:acme</id>
  <entry>
    <id>/myserver/v1/widgets/acme/205390.en.xml</id>
    <as:entryId>205390</as:entryId>
    <title type="text"> Entry: acme 205390.en</title>
    <author>
      <name>AtomServer Atom Service</name>
    </author>
    <link href="/myserver/v1/widgets/acme/205390.en.xml" rel="self" />
    <link href="/myserver/v1/widgets/acme/205390.en.xml/2" rel="edit" />
  </entry>
</feed>
*/

// Option 1 - AtomFormatter understands how to deal with it's own model. Feed, Entry, etc.
public class StoryController : ApiController
{
    public HttpResponseMessage Get(int id)
    {
        Story story = GetMeAStory(id);
        var response = new HttpResponseMessage(HttpStatusCode.OK);

        Entry entry = ConvertToEntry(story); // Why do we need this?
        response.Content = new ObjectContent<SyndicationItem>(entry, atomFormatter);
        return response;        
    }
}

// I'm not convinced we need media type specific models and translate twice. Atom example:
// Story (our model) -> Entry (atom model) -> Atom Representation
//
// I think we can do better.

// Option 2
var formatter = new AtomFormatter();
formatter.MapModel<Story>()
    .ToTitle(x => x.Topic)
    .ToAuthor(x => x.Owner);

// Option 2b - Since Atom is specifically XML, why not use it.
var formatter = new AtomFormatter();
formatter.MapModel<Story>()
    .To("/feed/title", x => x.Topic)
    .To("/feed/author/name", x => x.Owner);

// AtomFormatter.MapModel<T>() would return an AtomModelMap, so we still have SRP as the mapping and
// formatting are still independent.

public class StoryController : ApiController
{
    public HttpResponseMessage Get(int id)
    {
        Story story = GetMeAStory(id);
        var response = new HttpResponseMessage(HttpStatusCode.OK);
        response.Content = this.Request.Represent(story); // The proper formatter is used based on conneg
        return response;        
    }
}

// Option 2 and 2b don't require copying data for the purpose of mapping a translation like option 1 does.
// We've also eliminated a double translation that occurred in option 1.

Why not simply use the view-model pattern with various view rendering strategies ?

@DarkSmith feel free to fork the gist and create an alternative :)

I didn't like having an intermediate model just for the purpose of translation because that forces us into translating twice. I get to use multiple formatters for a given model and map to them without having to use a model in the middle. I don't believe copying data is required to do this representation.

Please fork the gist and show the alternative. Seeing the code will be easier to understand what you mean :)

Given that the available transitions/links are dependent of the resource state, I'm assuming the Story object would be responsible to know it's available transitions/links.

The formatter then would have to know how to retrieve the links from the story object in the response message. Is that correct?