I am designing some architectural changes into a legacy ASP.NET application. I prototyped some classes for dependency resolution that mimic the ASP.NET MVC's IDependencyResolver. I won't post because it is pretty much the same interface, but in other natural language.
I figured out it might be considered Service Location, which in turn is usually (not fully in some cases) condemned in favor of Dependency Injection. Nevertheless, I couldn't find any recommendation against the use of the ASP.NET MVC's dependency resolution implementation.
Is the ASP.NET MVC's IDependencyResolver considered an anti-pattern? Is it a bad thing?
If you look at the signature you will see that it's just a Service Locator with another name. Service Locator is an anti-pattern and I consider the relationship transitive, so I consider IDependencyResolver an anti-pattern.
Apart from that, the interface is also broken because it has no Release method.
I don't believe so... You can inject any IoC you want into ASP.NET MVC, which seems like a pretty good pattern to me.
Here's a blog post about injecting Unity into ASP.NET MVC 3.
Related
I recently finished reading Dino Esposito's great book Modern Web Development, and in it he addresses a suggestion for a Domain Driven Layered Architecture for web applications. I have always struggled with a specific piece of suggestions I have seen similar to the one below:
Specifically with reference to the IoC being made in the Infrastructure layer. I understand the reasoning behind this and it makes sense to me however how do you adequately implement that within the bounds of the ASP.NET MVC framework? To add a dependency resolver you need to implement the IDependencyResolver interface which exists in the System.Web.MVC namespace.
In past projects I would typically implement my IoC within the MVC application itself in the startup folder however this seems to be at odds with the suggestion for the layout.
I do not want to turn this into an opinion type of question, all I am looking for is a possible, actual concrete way to implement this pattern without dragging the System.Web.MVC namespace down to the infrastructure layer.
EDIT
To add a follow on diagram for the suggested architecture, and the part that is still confusing to me, it would appear that Dino's suggestion does indeed put the IoC container in the infrastructure assembly:
Answer to Your Question
Fundamentally, your question is "I am looking for is a possible, actual concrete way to implement this pattern without dragging the System.Web.MVC namespace down to the infrastructure layer"
There is a way to do this, and it involves introducing a new IoC container library, one dedicated for the purpose.
IDependencyResolver does not have to be your system wide resolution interface - it is just the interface used by MvC. There are other IoC containers, and a number of them provide adaptors to inject an implementation of IDependencyResolver that wraps their IoC logic.
This permits a few things:
The MvC components that depend on the ability to perform an explicit resolution can still depend on IDependencyResolver
Other layers in the system can depend on a different resolution interface, and thus contain a reference to an isolated fit-for-purpose assembly
Both the MvC layer and the other layers will all be accessing the same set of dependency/implementation registrations
Some examples of IoC containers that support this:
Autofac - with Autofac Mvc Support
You can see the last line of the sample is:
DependencyResolver.SetResolver(new AutofacDependencyResolver(container));
After that line, any MvC component that depends on IDependencyResolver will automatically get the AutofacDependencyResolver which wraps calls to the Autofac container
StructureMap - StructureMap.Mvc
Here is a comparison of a large number of c# IoC containers that may help you select the one that's right for you.
[Actual Implementation Concerns - aka My Opinion about Why this is NOT a good idea]
Your practice in your past projects of only using the IoC in the Mvc application is more correct, in my opinion, so the below concepts may already be familiar to you, but as you are considering referencing the IoC from the domain, I thought it worth exploring.
First question - Why?
While that answer provides a way to do what you're asking, based on that diagram, I confess it's not clear to me what the purpose is of depending on the IoC resolver from the domain layer, and why you would need to do that.
If you find yourself doing that, you may be accidentally using the Service Location Anti-Pattern
As outlined in that blog, there is no need to depend on the IoC resolver (or locator) - simply depend on the service you need, and let the IoC inject the appropriate implementation.
Part of the problem in understanding the intent is the diagram itself - it often happens that people draw diagrams by dropping on some boxes and connecting them up - without ever being clear about what the lines mean. Are they chains of dependency? Are they sequence of execution? What does it mean to have a line from the domain model box to the actual label of the infrastructure layer??? Is it depending on nothing? Or illustrative of a possible dependency that is not articulated here?
What should use the IoC resolver?
The only part of the system that should directly reference the IoC resolver is the composition root, which is effectively the entry point to the application. The first part 'wires up the object graph' - really, it registers how to resolve all possible dependencies from the interfaces that are depended on, to appropriate concrete implementations.
It then resolves the entry point object (or registers an IDependencyResolver so Mvc can resolve the entry point object, aka a controller). When the entry object is resolved, it automatically resolves all it's dependencies, in the process resolving next layer of dependencies, and so on all the way until you reach classes with no dependencies. Which is likely to be your domain layer, if you are doing DDD.
Dependency-less Domain Layer and the Onion Architecture
Since you are interested in DDD, the received wisdom is that the domain layer should not depend on anything that is not defined in the domain layer. If there is really a need to utilise the services of an infrastructure component such as a repository, use separated interfaces and put the interface in the domain layer, but the implementation in a concrete persistence layer.
The architectural pattern this lends itself to is known as the Onion Architecture also known as the Hexagonal Architecture
Using Other IoC Containers
While I don't think it's necessary to reference the IoC resolver/locator from the domain layer (or any layer, really), I do still think there is value in adopting a separate dedicated IoC container library, as outlined above.
The value is in some of the more flexible options for how to configure services, including some nifty convention based auto-configuration.
The one reason it might be worth depending on the IoC library in the domain layer is to co-locate the registration and configuration logic with the services that are being configured, which can help structure and organise your IoC dependency registrations. But just because you take a dependency on the IoC assembly to permit structuring your registrations, doesn't mean you should use the IoC resolver/locator.
Recently I came across this discussion.
In ASP.NET WebAPI, when we need to integrate a custom inversion of control/dependency injection container, we need to go with two different strategies:
Implement IHttpControllerActivator, which is an extension point to fully control controller's life-cycle. That is, we can define how a controller is instantiated. At this point, we can resolve the controller using an inversion of control container and let it use its own dependency injection approach.
Implement IDependencyResolver, where we can define how to resolve dependency injections.
In my current projects I went the IHttpControllerActivator way, and since I'm using Castle Windsor as inversion of control container, I can fully control object life cycle from its container configuration, and I can decide how controllers are resolved, and the scope of its recursively-injected dependencies defining that they die once the controller ends its life (i.e. when the request ends).
While some of these features can be achieved with IDependencyResolver implementations, I feel that IHttpControllerActivator is the best way to integrate inversion of control and dependency injection in ASP.NET Web API, because I believe that I prefer to stay as abstract as possible and stick with Castle Windsor configuration model in terms of inversion of control and dependency injection.
The main drawback with the IHttpControllerActivator approach is you need to register all controllers in the container while IDependencyResolver still gives the responsibility of resolving the controller to ASP.NET Web API pipeline (for me, this hasn't been a big issue, I just used Castle Windsor's Classes.FromAssembly to configure all controllers deriving ApiController).
Do I ignore any other drawbacks that would make the IDependencyResolver approach more appropiate?
The IHttpControllerActivator approach primarily has an advantage over the IDependencyResolver approach because of the context provided, as excellently described in the following blog post: http://blog.ploeh.dk/2012/09/28/DependencyInjectionandLifetimeManagementwithASP.NETWebAPI/.
However, in most easier scenarios, it simply does not pay back the extra effort to manually register everything. Using the IDependencyResolver approach in many situations just 'does the trick'. Add good documentation and the larger day to day user base to that, and I myself would probably consider if I could manage using IDependencyResolver before considering IHttpControllerActivator.
I'm moving from StructureMap 2.x to 3.x. One major change is that using ObjectFactory results in the following warning:
'StructureMap.ObjectFactory' is obsolete: 'ObjectFactory
will be removed in a future 4.0 release of StructureMap. Favor the
usage of the Container class for future
work'
So in most cases, the resolution is fairly easy: pass through IContainer as a constructor. Unfortunately this is not feasible for ASMX web serivces or attributes, both of which require a default constructor. That means I'm probably stuck with the Service Locator Pattern, property injection, or writing my own ObjectFactory implementation.
What's the preferred way to deal with this unfortunate problem?
Edit: It's worth mentioning that my container does an Assembly scan.
Per Jeremy Miller, himself:
Assembly scanning isn’t cheap and you (almost?) always wanna cache the
results. So, yeah, in that case you’re going to have to write your
own ObjectFactory. Someday all that bad old MS tech will go away.
So in this case, this implementation is what should be done.
The cleanest way I have seen to deal with this is to use .NET routing to take control of the entry point, then make a custom PageHandlerFactory implementation that takes the DI container as a dependency.
The custom page handler factory will then property inject the page/service after it is instantiated but before any of its events are called.
This is almost exactly the same way that IControllerFactory works with DI in MVC. In MVC, the container is injected into a custom IControllerFactory implementation at application startup, which effectively makes it a part of the application's composition root. In the case of ASP.NET, the IRouteHandler would effectively be part of the composition root.
I was unable to find a link to the source where I originally saw that implementation. But this one is pretty close. The primary difference is that one tries to use constructor injection, but the downside is that it requires full trust to do. I believe if you stick with property injection, you can do it with partial trust.
I need to fetch data from an external API, only accessible via VPN.
The development/test machine will not always be able to connect to the VPN.
The desired behaviour is to use two different implementations (one that calls the actual external API and one that acts as the real thing but returns dummy data). Which implementation to use will be configured via a flag in web.config
I've tried the IoC containers StructureMap and Unity and they both did the job but they only seem to be applicable for MVC, I'm looking for a generic solution that also works for web forms. And also, isn't it a bit overkill to use them for this isolated design problem!?
Is there a design pattern or best practice approach for this particular scenario?
IoC / dependency injection sounds like the correct approach, but you don't necessarily need a container for a simple scenario. The key is to have classes that depend on the API reference an interface IAPI, and pass it the actual implementation RealAPI or FakeAPI.
public class SomeClass
{
private readonly IAPI _api;
public SomeClass(IAPI api)
{
_api = api;
}
}
Now you should be able to switch out the implementation easily by passing a different object to MyClass. In theory, when you're using an IoC approach, you should only need to bind the interface to the implementation once, at the top level of the application.
isn't it a bit overkill to use them for this isolated design problem!?
They probably are. Those IoC containers only help you when you wrote loosly coupled code. If you didn't design your classes according to the SOLID principles for instance, those frameworks will probably only be in the way. On the other hand, which developer doesn't want to write loosly coupled code? In other words, IoC container solves a problem you might not have but it's a nice problem to have.
StructureMap and Unity [...] only seem to be applicable for MVC
Those ioc frameworks can be used in any type of application (as long as it is written in loosly coupled way). Some types of applications need a bit more work to plug a framework in, but it's always possible. StructureMap and Unity might only have integration packages for MVC, it's quite easy to use them in ASP.NET Web Forms as well.
Is there a design pattern or best practice approach for this
particular scenario?
What you're looking for is the Proxy pattern and perhaps the circuit breaker pattern.
I've used MEF for a few projects at work and I've just stared messing about with Monorail in my spare time. I was wondering if there was any way that I could use MEF to load the controllers that Monorail uses. Monorail appears to look for controllers in assemblies that you list in the Web.Config:
<controllers>
<assembly>my.assembly</assembly>
</controllers>
Is there a way that 'my.assembly' can then use MEF to load up more controllers? I have to admit I haven't though of a reason I would need this functionality but I'm just trying things out!
Monorail is a very extensible framework. Almost everything is being provided by a service that can be easily switched with something else.
For e.g., IoC integration is quite easy, as you can switch the services that creates controllers, filters, helpers, and all other MonoRail entities.
Concrete example: Integrating Windsor container into Monorail
Now this sets up almost everything in the Monorail to be provided by Windsor. If you only want Controllers to be provided by MEF, there's even less work.
I have very little working knowledge of MEF so it might be a little off, but you'd get the general idea:
Use MEF discovery mechanisms to locate controller types, then add controller types to the default IControllerTree service. take a peek at MonoRailFacility.cs for inspiration.
Implement a MefControllerFactory : IControllerFactory that will use MEF to instantiate controllers when needed. Inspiration is at WindsorControllerFactory.cs
It's not something you can use yet, but have a look at Hammett's blog post here, where he talks about what he is prototyping for MonoRail 3.0 (including support for MEF by default).