ServiceStack zero dependency Request-Response DTOs - dependency-injection

After reading some ServiceStack wiki, I have a problem about DTO and I was hoping you could help.
The wiki said:
In Service development your services DTOs provides your technology agnostic Service Layer which you want to keep clean and as 'dependency-free' as possible for maximum accessibility and potential re-use. Our recommendation is to keep your service DTOs in a separate largely dep-free assembly.
(https://github.com/ServiceStack/ServiceStack/wiki/New-API)
Finally you can also use the previous more explicit client API (ideal for when you don't have the IReturn<> marker):
(https://github.com/ServiceStack/ServiceStack/wiki/New-API)
According to the reasons above, I consider the best practice about ServiceStack is:
We should use POCO Request-Response DTOs instead of inheriting from IReturn<>.?
For instance:
We should use 1#:
public class AuthenticationRequest
{
public string Name { get; set; }
public string Password { get; set; }
}
public class AuthenticationResponse
{
public AuthenticationResponseType Result { get; set; }
public UserInfoDto UserInfo { get; set; }
}
We shouldn't use 2#:
using ServiceStack;
public class AuthenticationRequest : IReturn<AuthenticationResponse>
{
public string Name { get; set; }
public string Password { get; set; }
}
public class AuthenticationResponse
{
public AuthenticationResponseType Result { get; set; }
public UserInfoDto UserInfo { get; set; }
}
Because 1# is zero dependency, 2# have a dependency on ServiceStack library/framework.
If I package all Request-Response DTOs to a NET DLL, 1# is more abstract than 2#!
This means:
If one day in the future I deceide not to use ServiceStack, this DLL doesn't need any change. (ServiceStack library/framework should be Infrastructure not Abstraction)
Please correct me if I am wrong.
Very thanks.

The only dependency DTO's should have is the impl-free ServiceStack.Interfaces.dll which as it's a Portable Class Library (PCL) supports almost every mobile or Desktop platform that .NET runs on. ServiceStack's Interfaces .dll is required in order to be able to cleanly describe your complete Services contract in a single, benign .dll.
For example. the [Route] metadata attribute captures the Custom Routes where the remote Services are hosted which is required info about your Service that clients need to know in order to be able to call services via their published Custom Routes. Likewise the IReturn<T> interface marker provides a strong-typed contract on what your Service returns which is what enables ServiceStack succinct end-to-end Typed API. Essentially ServiceStack.Interfaces is a required extension to be able to capture your entire Service Contract in your Services DTO's.
ServiceStack.Interfaces can be used outside of ServiceStack
Even if you don't use ServiceStack, you can still use the benign ServiceStack.Interfaces.dll which the clients can introspect to find out more information about your DTO's and the remote Service Contract. Whilst I'm not seeing any reason to, if you want to decouple the ServiceStack.Interfaces on your project you can just copy the attributes you're using in your DTO .dll freeing it from any external dependencies. But this would impact your ability to have a generic Service Client since these embedded interfaces and attributes are unknown to your client library, limiting its ability to enable rich generic functionality using it.
Service Contract Interfaces and Attributes in other Languages
To support non .NET languages like TypeScript, ServiceStack emits these interfaces in the generated DTO's so they don't require any dependencies.
Likewise in Add ServiceStack Reference support of Swift 2.0 or Java and Android these additional contracts are emitted idiomatically referencing a Swift IReturn protocol or IReturn<T> interface in the Java android client package which is also what enables the succinct Typed API's ServiceStack enables on both iOS and Android.
Service Design
Something you should keep in mind when designing your API's is that your Service Layer is your most important contract. i.e. Your API exists to allow consumers access to your remote Servers capabilities, so your internal logic should be a hidden impl-detail, not something that should impact the external surface area of your API.
The Request DTO defines your Service Contract where I find using a Request suffix is an ugly construct that negatively affects the readability of your external API, e.g. Here's a typical example of what a noun with a *Request suffix would look like:
var response = client.Get(new CustomerRequest { ... });
Compared with using a Verb where the Request DTO is indicative and provides better readability of what the Service does:
var response = client.Get(new FindCustomers { ... });
Your Request DTO should ideally be a verb that's grouped by call semantics and Response Type. Having a *Dto suffix is an indication that your internal implementation is leaking and affecting the ideal Service Contract your external API Consumers will bind to (and should never change). Keep in mind the objective of your Service is to provide re-usable functionality to your consumers so your impl should realize your published contract, not the other way around where its implementation dictates what the contract should be.
With that in mind I would rewrite your ServiceStack Examples to look like:
public class Authenticate : IReturn<AuthenticateResponse>
{
public string UserName { get; set; }
public string Password { get; set; }
}
public class AuthenticateResponse
{
public AuthenticationResult Result { get; set; }
public UserInfo UserInfo { get; set; }
}
Which ends up being similar to ServiceStack's built-in Authenticate and AuthenticateResponse Request and Response DTOs.
I also recommend reading this earlier answer to understand the importance of DTO's and how it relates to the goals of a Service.

Related

Access HttpConfiguration.DependencyResolver in validation attribute

I have a web api application with Autofac. For the input insert models I need to use validation attributes for the properties that indicate related entities.
public class Comment
{
[Required]
[ExistentBookValidationAttribute]
public int BookId { get; set; }
}
In ExistentBookValidationAttribute I need to access a business service to do the validation. Since Autofac doesn't inject properties to the validation attributes I decided to use the dependency resolver to get the service manually. But I don't want to use GlobalConfiguration.Configuration.DependencyResolver. I'd like to use DependencyResolver from web api HttpConfiguration. So is that possible? Is HttpConfiguration.DependencyResolver accessable in the validation attributes?

WebAPI ODATA -- Consuming With MVC Using Standard Controller/Ioc/Repository/UoW Sort of Architecture

Working with WebAPI ODATA services with javascript is not a problem... but what is a current recommendation to wrap the http calls (CRUD) to be consumed through a MVC5 application with a repository. Much of the guidance I see ultimately goes directly to the entity/dbcontext. I am looking for guidance which demonstrates the "drinking of your own Kool-Aid" and consuming the same ODATA (and it can be plain WebAPI, also) published externally to consumers of an application.
In my mind, I'm looking at this sort of flow:
AppController (site1:443)-->AppRepository-->OdataController (apiSite2:443)-->OdataRepository-->DataSource
The secondary concern is that I don't necessarily want direct access to a datasource by any consumer--especially posts without being vetted and I don't want all (any) of the logic in the controller. I might be overthinking something...
In order to extract the business logic from the controller I typically either push said logic down to domain objects whenever possible. If that isn't possible, then I'll create a class specifically designed to manage the logic in question, such as an interaction between two different objects.
If all else fails, then I'll have the interaction managed by a service. The classes might look something like the following:
public class SomeApiController : ApiController
{
public SomeApiController(ISomeApiService service)
{
this.Service = service;
}
private ISomeApiService Service { get; set; }
public IHttpActionResult SomeMethod(int someObjectId)
{
// service manages the logic and either defers to the object in question or resolves it through some specialized class
var result = this.Service.SomeMethod(someObjectId);
return this.OK(result);
}
}
public class SomeApiService : ISomeApiService
{
public SomeApiService(ISomeRepository repository)
{
this.Repository = repository;
}
private ISomeRepository Repository { get; set; }
}
... and so on.
The idea being that the layers have no dependencies upon one another which cannot be resolved through the IoC container of your choice and that the dependencies only go one way. That is to say SomeApiService has no dependency on SomeApiController and SomeApiRepository would have no dependency on SomeApiService.

where to keep frequently used methods in MVC

I need to implement MVC architecture in my company, So can anyone suggest where to keep frequently used methods to call on all pages. Like:
states ddl, departments ddl also roles list and etc...
Please give me suggestions where to keep them in architecture.
Thanks
There are different solutions depending on the scale of your application. For small projects, you can simply create a set of classes in MVC application itself. Just create a Utils folder and a DropDownLists class and away you go. For simple stuff like this, I find it's acceptable to have static methods that return the data, lists, or enumerations you require.
Another option is to create an abstract MyControllerBase class that descends from Controller and put your cross-cutting concerns in there, perhaps as virtual methods or properties. Then all your actual controllers can descend from MyControllerBase.
For larger applications, or in situations where you might share these classes with other MVC applications, create a shared library such as MySolution.Utils and reference the library from all projects as required.
Yet another possibility for larger solutions is to use Dependency Injection to inject the requirements in at runtime. You might consider using something like Unity or Ninject for this task.
Example, as per your request (also in GitHub Gist)
// declare these in a shared library
public interface ILookupDataProvider
{
IEnumerable<string> States { get; }
}
public class LookupDataProvider: ILookupDataProvider
{
public IEnumerable<string> States
{
get
{
return new string[] { "A", "B", "C" };
}
}
}
// then inject the requirement in to your controller
// in this example, the [Dependency] attribute comes from Unity (other DI containers are available!)
public class MyController : Controller
{
[Dependency]
public ILookupDataProvider LookupDataProvider { get; set; }
public ActionResult Index()
{
var myModel = new MyModel
{
States = LookupDataProvider.States
};
return View(myModel);
}
}
In the code above, you'll need to configure your Dependency Injection technology but this is definitely outside the scope of the answer (check SO for help here). Once configured correctly, the concrete implementation of ILookupDataProvider will be injected in at runtime to provide the data.
One final solution I would suggest, albeit this would be very much overkill for small projects would be to host shared services in a WCF service layer. This allows parts of your application to be separated out in to highly-scalable services, should the need arise in the future.

How dependant is breezjs serverside contexprovider to entityframework?

I am building a WebApi for a CMS that has its own data provider. No DBContext or entity framework involved.
I have previously used breeze as it being such a breeze to map the server side model to the client:)
I have wondering if I can extend my code or breeze in a way such I get all the stuff from brezejs for free.
What I have to work with is the following Interfaces that I have made implementations for based on the data provider from the CMS.
public interface IC1Repository<T>
{
IQueryable<T> GetAll();
T Add(T item);
void Remove(T item);
bool Update(T item);
}
Its generic, so thats not going to work on the client.
I can generate a context class i guess that holds all the types exposed.
public class mycontext
{
public IC1Repository<Category> Categories { get; set; }
public IC1Repository<Customer> Customers { get; set; }
public IC1Repository<Employee> Employees { get; set; }
}
What would my next steps be to get this workign with breeze. Are there any interfaces i can implement such it mimics the DbContext. Can i maybe crate my custom DbSet that do not talk with a database, but just is a implementation of my IC1Repository above?
Any advices thanks :)
I think you want the ContextProvider which is the base class of the EFContextProvider.
That has the same semantics and same base behavior as the EFContextProvider but it doesn't use EF.
Check out the "No DB" sample which uses the ContextProvider to manage queries and saves to an in-memory "database".
Ignore the fact that this class sits in a DLL with references to EF. I realize that is annoying. But your project will compile and run just fine when there are no EF assemblies around. You can delete all the EF stuff if you used NuGet to get the Breeze.WebApi.dll.

Validation approach on MVC3 CRUD application with EF4-based WCF as backend

I develop a simple MVC3 CRUD application - simple controllers / views, which uses WCF service for CRUD data access.
The WCF uses EF4.1 with DbContext, and simple CRUD-style methods: ListEntities, GetEntity(ID), AddEntity (entity), DeleteEntity(ID)
If I develop the MVC application directly with EF, code first, I can annotate properties in the entity classes with validation attributes, and the MVC application will automatically recognize validation errors and report them in the UI when I try to save and a validation error occurs (e.g. a required field is not set).
But in my application I don't use this approach and I face two problems:
My entities in the WCF are generated from the EDMX, which in turn was also generated from the database. So I cannot actually add to them any data validation annotation attributes, because they'll vanish as soon as the entities will be regenerated from the EDMX. Is there any solution to this?
Since my client (MVC app) does not share the data contract classes with WCF (for clear separation), but instead it is generated form service reference, even if I find a way to add data annotation attributes to server-side data contract classes, will they be recognized and recreated when the data contract proxy class is created on client side?
So how could I made the MVC application to use client side validation and error message reporting for validation failures when binding to entities exposed by WCF service as data contracts?
One idea I have is, on client side, to create derived classes for all entities exposed as data contracts, and apply annotation attributes to them to desired properties. But this doesn't looks like a good solution to me, because with this I create a logic "coupling" between UI client and the WCF service / data layer (forcing UI to know about data more than it should do - by putting BL logic in client).
Can anyone give me some suggestions on how to handle those this situation?
Thanks
1: Yes you can add validation using the System.ComponentModel.DataAnnotations.MetaDataType.
I answered this question at MVC Partial Model Updates
2a: What you can do is create a seperate Class Library Assembly that contains all the interfaces (with or without additional MetaDataTypes) and use that on both the WCF service and the MVC application. After you add the reference to your MVC application, when adding the WCF Service reference, you can match the WCF Service DataContacts directly to the interfaces in the Assembly. One Caveat is that both the WCF service and MVC application are dependant on the Assembly (some might consider this tightly coupled) but this should be ok because you are only tightly coupling at the interface level, and whether or not you choose to allow VS to recreate it's own interfaces/classes or reuse what you already created in the Assembly it boils down to the same thing in my opinion.
2b: If you decide not to use a Class Library, I'm pretty sure that the service reference classes are partial, and you can simply create another .cs file with partial classes and add the interfaces as I described in part 1 to the partial classes.
Update
I am currently using Entity Framework to access my database. Entity Framework, like WCF References, classes are Auto-Generated classes will look something similar to:
[EdmEntityTypeAttribute(NamespaceName="MyNameSpace", Name="Info ")]
[Serializable()]
[DataContractAttribute(IsReference=true)]
public partial class Info : EntityObject
{
public static Info CreateInfo (global::System.Int32 id)
{
Info info= new Info ();
info.Id = id;
return info;
}
public string Name { get; set; }
public string FavoriteColor { get; set; }
// etc etc
}
In a separate file with the same namespace as the previous partial class, I have created:
[SomeAttribute1]
[AnotherAttribute2]
public partial class Info: IInfo
{
}
So now my auto-generated class is not only based on an Interface I created IInfo so the actual methods are not exposed (because my datatier in MVC returns interfaces), but it also has Attributes (for Data Annotations or whatever).
What I would suggest is instead of putting your data annotations directly on your WCF Service reference class is to use the MetedataType DataAnnotations. This allows you to separate the actual data object with the data annotations validations. Especially helpful if you want to use the same data class with different validations based on whatever (maybe administrators don't have to have a valid favorite color).
For example:
public interface NormalUser
{
[Required]
string Name { get; set; }
[Required]
string FavoriteColor { get; set; }
}
public interface AdminUser
{
[Required]
string Name { get; set; }
string FavoriteColor { get; set; }
}
[MetadataType(typeof(INormalUser))
public class NormalUserInfo : Info { }
[MetadataType(typeof(IAdminUser))
public class AdminUserInfo : Info { }
In this example we have two different classes NormaUserInfo and AdminUserInfo which both have different validations. Each of them have inherited from Info so they are valid models that can be passed into the WCF Service.
Out of my mind, as I can't test it right now...
Let's say your autogenerated code is like this:
public partial class Employee
{
//some code here
}
You can add a new Employee class, also partial, and this one won't be autogenerated
[you can annotate here]
public partial class Employee
{
//somecode here
}
try it
As for the validation, you could use: http://fluentvalidation.codeplex.com/

Resources