How do I manage object disposal when I use IoC? - dependency-injection

My case it is Ninject 2.
// normal explicit dispose
using (var dc = new EFContext)
{
}
But sometimes I need to keep the context longer or between function calls.
So I want to control this behavior through IoC scope.
// if i use this way. how do i make sure object is disposed.
var dc = ninject.Get<IContext>()
// i cannot use this since the scope can change to singleton. right ??
using (var dc = ninject.Get<IContext>())
{
}
Sample scopes
Container.Bind<IContext>().To<EFContext>().InSingletonScope();
// OR
Container.Bind<IContext>().To<EFContext>().InRequestScope();

From what I know (I did a research about a month ago) Ninject does not support lifecycle management at all. Castle Windsor and AutoFac (and to some extent StructureMap, but only when using nested containers) will take care of disposing disposable components they create at appropriate time.

If you have control over the interface of IContext, add IDisposable to the list of interfaces from which it inherits. If not, downcast the IContext you get to an IDisposable...
var context = ninject.Get<IContext>();
using ((IDisposable)context)
{
}
You also have the option of altering the interface of IContext to do this by composition, if you control IContext...
public interface IContext
{
// ...
IDisposable GetUsageHandle();
}
var context = ninject.Get<IContext>();
using (context.GetUsageHandle())
{
}

In addition to the standard scopes of Transient, OnePerThread, and Singleton, you can use an ActivationBlock in order to control the lifetime of a whole set of objects. When the block is disposed, all object retrieved by the block go out of scope - so singletons and others are disposed of when their instances are requested by the activation block.
var kernel = new StandardKernel();
kernel.Bind<NotifiesWhenDisposed>().ToSelf();
NotifiesWhenDisposed instance = null;
using(var block = new ActivationBlock(kernel))
{
instance = block.Get<NotifiesWhenDisposed>();
instance.IsDisposed.ShouldBeFalse();
}
instance.IsDisposed.ShouldBeTrue();

Related

Are Factories Using An IoC Container A Service Locator?

Lets say I have a factory returning different classes via methods.
class CarFactory
{
public function __construct(ContainerInterface $container)
{
$this->container = $container;
}
public function createCarOne() : CarInterface
{
return $this->container->make(CarOneClass::class);
}
// Vs
public function createCarTwo() : CarInterface
{
return new CarTwoClass({Inject Dependencies Here});
}
}
When would this be considered a service locator or anti-pattern and why? I am considering the first method solely for the dependency resolution provided by the container. All car's have the same typed interface dependencies the main difference of the entities come from how they transform the data provided.
Whenever one of these methods are called I need a new instance of the specified car so the data set can be transformed based on the choice.
This is not the implementation but the easiest example I can provide.
$output = [];
foreach ($car as $key => $data) {
$newCar = $this->factory->createCar{$key}();
// Pass Some Data To The New Car Methods So It Can Be Transformed
$output[] = $newCar;
}
return $output;
If this is the wrong approach what would be the alternative option?
Edit
After further digging I see some IoC containers pass factory callables as dependencies. I was going to bind each Car to a callable but thanks to the ability to type hint data from method returns (php7) I can configure factories using a provider then call the 'callable factory' from within the CarFactory. Requires additional binding but prevents the need to reference/dependency inject the IoC container within every factory.
Still researching I would love to hear feedback from those with more experience.
Ex:
// Within Some Registered Provider
// I Will Have To Wire Each Car
$one = function() use ($app) {
return $app->make(CarOne::class);
};
$two = function() use ($app) {
return $app->make(CarTwo::class);
};
$app->bind(ICarFactory::class, function($app) use ($one, $two) {
return $app->make($concrete, [$one, $two]);
});
// Car Factory Constructor
public function __construct(callable $carOne, callable $carTwo) {
$this->one = $carOne;
$this->two = $carTwo;
}
Since get methods are type hinted ( view original car factory ) an error is thrown when the returned item does not implement CarInterface, each factory method would just have to call the 'callable factory' ( something like this return ($this->one)();).
I believe i solve my problem of outsourcing creation of dependencies ( avoiding creating within factory was bothering the hell out of me ) while still following 'best practices'. Still looking for advice if anyone has any to offer.

Setting Session on Specific Event Calls

I have an MVC application and a custom class called AuthorisationFilter which has a .NET interface of IAuthorizationFilter, this has an OnAuthorization method which gets called when I click around my site, at that point I go about validating the security access of the user (which works), but I don't want to do this all the time as it is time consuming.
In this I'm trying to use the Session to store a temporary piece of login information (this is an internal application by the way), but I can't get it working as I'd expect. I can't just use an HttpContext so end up constantly creating a new instance of HttpContextBase, which I assume is then clearing out the Session. My code is as follows:
internal void SetSecurityLevel(int token)
{
HttpContextBase _cBase = new HttpContextWrapper(HttpContext.Current);
_cBase.Session["SecurityRights"] = token;
}
internal int GetSecurityLevel()
{
HttpContextBase _cBase = new HttpContextWrapper(HttpContext.Current);
if (_cBase.Session["SecurityRights"] == null)
{
SetSecurityLevel(-1);
}
return (int)_cBase.Session["SecurityRights"];
}
Please note this is only part of the code, SetSecurityLevel is set to the correct value by a separate method call which is not shown
Anyway what I'm really wanting to do is have the session set in this class and have it persisted. I tried a few different ways, including setting the context when the class is initialised, but I end up with a NullReference on the .Session object in GetSecurityLevel
private HttpContextBase _cBase = new HttpContextWrapper(HttpContext.Current);
public AuthorisationFilter()
{
_cBase = new HttpContextWrapper(HttpContext.Current);
}
Is there a way I can do this within the class?
You are right that you can't set in the constructor but you can in OnActionExecuting when the context is available
public override void OnActionExecuting(ActionExecutedContext filterContext)
{
_session = = filterContext.HttpContext.Session;
I would wonder a little why you're setting this security level in the filter and not just doing it directly from where it's needed.

Autofac: any way to resolve the innermost scope?

I'm currently trying out Autofac in a new ASP.NET MVC project after having used Ninject, Castle Windsor and other IoC containers in the last years. So while I know about IoC containers in general, I'm fairly new to Autofac and I'm still looking for some best practices.
Currently I'm trying to find out if there is a way to resolve the innermost nested scope.
I have the following situation: a component that is registered as SingleInstance() has a method that creates a nested lifetime scope, providing a configuration action to configure some components as InstancePerLifetimeScope, and within this nested scope resolves the registered components to do something useful, like so:
ILifetimeScope currentScope = ???;
using (var scope = currentScope.BeginLifetimeScope(cb => {
cb.RegisterType<X>().InstancePerLifetimeScope();
// ...
}))
{
var comp = scope.Resolve<X>();
// ...
}
The issue is that I would like currentScope to be the innermost lifetime scope, because I know that X depends on components inside the innermost scope. In the simplest case that would be e.g. the current request lifetime scope. I can of course get it with AutofacDependencyResolver.Current.RequestLifetimeScope but I don't want to use that as it isn't really well testable. Also, that lifetime scope isn't necessarily the innermost.
So, is there a way to find the innermost lifetime scope given e.g. the root container or a different ILifetimeScope?
In Autofac, the innermost scope is always the container. Using the AutofacDependencyResolver, it'd be
AutofacDependencyResolver.Current.ApplicationContainer
There is no way from a nested scope (if all you have is an ILifetimeScope) to "walk backward" to get to the container. I'm not necessarily sure you want to do that, anyway.
It sounds like your SingleInstance component is doing some sort of service location, basically, with manual registration/resolution of certain components. If the set of types being registered is fixed, I might recommend (if possible) some redesign of your system, so the SingleInstance component isn't registered as SingleInstance anymore and instead gets registered as InstancePerDependency, then have that take these other items in as constructor parameters.
Instead of...
// Consuming class like this...
public class BigComponent
{
public void DoSomethingCool()
{
using(var scope = ...)
{
var c = scope.Resolve<SubComponent>();
c.DoWork();
}
}
}
// ...and container registrations like this...
builder.RegisterType<BigComponent>().SingleInstance();
You might try inverting it a bit:
// Consuming class like this...
public class BigComponent
{
private SubComponent _c;
public BigComponent(SubComponent c)
{
_c = c;
}
public void DoSomethingCool()
{
_c.DoWork();
}
}
// ...and container registrations like this...
builder.RegisterType<BigComponent>().InstancePerDependency();
builder.RegisterType<SubComponent>().InstancePerLifetimeScope();
The idea is to not have to do the on-the-fly registration-and-immediate-resolution thing.
If you're stuck doing service location, you'll need to use AutofacDependencyResolver.Current.ApplicationContainer if you need the absolute innermost scope, but keep in mind any objects you register scoped to InstancePerHttpRequest will not be resolvable if you do that, so you could get into trouble. It really is recommended to use the AutofacDependencyResolver.Current.RequestLifetimeScope instead. That would make your method:
var requestScope = AutofacDependencyResolver.Current.RequestLifetimeScope;
using (var scope = requestScope.BeginLifetimeScope(cb => {
cb.RegisterType<X>().InstancePerLifetimeScope();
// ...
}))
{
var comp = scope.Resolve<X>();
// ...
}
In a testing environment, the AutofacDependencyResolver lets you swap in the provider that dictates how request lifetimes get generated. You can implement a simple/stub one like this:
public class TestLifetimeScopeProvider : ILifetimeScopeProvider
{
readonly ILifetimeScope _container;
private ILifetimeScope _lifetimeScope = null;
public TestLifetimeScopeProvider(ILifetimeScope container)
{
if (container == null) throw new ArgumentNullException("container");
_container = container;
}
public ILifetimeScope ApplicationContainer
{
get { return _container; }
}
public ILifetimeScope GetLifetimeScope()
{
if (_lifetimeScope == null)
{
_lifetimeScope = ApplicationContainer.BeginLifetimeScope("httpRequest")
}
return _lifetimeScope;
}
public void EndLifetimeScope()
{
if (_lifetimeScope != null)
_lifetimeScope.Dispose();
}
}
Again, just a stub for unit testing, not something you'd ever use in production.
Then when you wire up the DependencyResolver in your test, you provide your lifetime scope provider:
var lsProvider = new TestLifetimeScopeProvider(container);
var resolver = new AutofacDependencyResolver(container, lsProvider);
DependencyResolver.SetResolver(resolver);
This lets you use InstancePerHttpRequest and such inside unit tests without actually having a real request context. It also means you should be able to use the request lifetime scope in your registration/resolution method and not have to fall back on the application container.
For those who are searching for ASP.NET WebApi:
You can use GetRequestLifetimeScope() method of AutofacWebApiDependencyResolver.

Storing context object in Request object, is it disposed of?

I am writing a MVC3 application, using NInject DI and repository pattern. Ninject is set up so that the repositories have a per-request lifetime.
I am putting the context object into the Http Request object, using the following code:
public static MessengerEntities GetContext()
{
if (!HttpContext.Current.Items.Contains("_db_context"))
{
HttpContext.Current.Items.Add("_db_context", new MessengerEntities());
}
return (MessengerEntities)HttpContext.Current.Items["_db_context"];
}
Then each repository calls this procedure to get either an existing or a new context object, e.g.:
public class TestRepository : ITestRepository
{
private MessengerEntities context = ContextHelper.GetContext();
#region ITestRepository Members
private string _testProperty = "blah";
public string testProperty
{
get
{
_testProperty = context.UserLogins.Where(n => n.inactive == null || !n.inactive.Value).ToList().Count.ToString();
return _testProperty;
}
set
{
_testProperty = value;
}
}
#endregion
}
(Later on, I plan to use a generic IRepository pattern, but for now I am just using this test repository.)
My question is: when the Request object is disposed of, will it also dispose of the context object in the Items collection? In other words, will it call Dispose on each object that may be stored in that collection?
I know there are a lot of discussions about this issue here, but they all seem to involve scenarios that are not quite the same as mine, so it's kind of hard to divine the answer.

Using WCF DataContract in MVC SessionState using AppFabric cache

I have a Data Access Layer, a Service Layer, and a Presentation Layer. The Presentation Layer is ASP.NET MVC2 RTM (web), and the Service Layer is WCF (services). It's all .NET 3.5 SP1.
The problem is that in the services, the objects being returned are marked with the [DataContract] attribute. The web is using the AppFabric Cache (a.k.a Velocity) SessionStateProvider to store session state. Due to this, anything I store in the session must be serializable.
Here comes the problem: the DataContracts aren't marked with [Serializable] and as far as I can remember, by introducing it onto a class already marked with [DataContract] some issues arise, and so I don't believe this is a solution.
I was initially planning on using the DataContracts right in the web layer, using them as models to views related to rendering the DataContracts (probably nested inside a higher level ViewModel class). But due to the session state provider requiring all objects stored inside it to be serializable, I'm starting to rethink this strategy. It would be nice to have though, since they contain validation logic using the IDataErrorInfo interface, and the same validation logic could be re-used in MVC as part of model binding.
What do you believe is the best way to allow me to reduce the work needed?
I've currently thought of the following different ways:
A. Create a 'ServiceIntegration' part in the web project.
This would be a middle man between my controllers and my WCF service layer. The ServiceIntegration part would speak to the service layer using DataContracts, and to the Web layer using ViewModels, but would have to transform between the DataContracts and ViewModels using a two-way Transformer.
Also, since the IDataErrorInfo Validation wouldn't be re-usable, it would be necessary to create a Validator per DataContract too, that uses the Transformer to convert from ViewModel to DataContract, perform validation using IDataErrorInfo and return its results. This would then be used inside action methods of Controllers (e.g. if (!MyValidator.IsValid(viewModel)) return View();)
Different classes required: xDataContract, xViewModel, xTransformer, xValidator
B. Create a 'SessionIntegration' part in the web project
This would be a middle-man between the controllers (or anything accessing the session) and the session itself. Anything requiring access to the session would go through this class. DataContracts would be used in the entire application, unless they are being stored into the session. The SessionIntegration part would take the responsibility of transforming the DataContract to some ISerializable form, and back. No additional Validator is needed because of the use of of IDataErrorInfo interface on the DataContract.
Different classes required: xDataContract, xTransformer, xSerializableForm
Note: there would still be ViewModels around in both scenarios, however with (B) I'd be able to compose ViewModels from DataContracts.
(B) has the benefit of not needing an extra validator.
Before I go off and implement (A)/(B) fully, I'd like some feedback. At the moment, I'm starting to lean towards (B), however, (A) might be more flexible. Either way, it seems like way too much work for what it's worth. Has anyone else come across this problem, do you agree/disagree with me, and/or do you have any other way of solving the problem?
Thanks,
James
Without going the full blown route of A or B, could you just make a generic ISerializable wrapper object and put those in your SessionState?
[Serializable]
public class Wrapper : ISerializable
{
public object Value { get; set; }
void ISerializable.GetObjectData(SerializationInfo info, StreamingContext context)
{
if (Value != null)
{
info.AddValue("IsNull", false);
if (Value.GetType().GetCustomAttributes(typeof(DataContractAttribute), false).Length == 1)
{
using (var ms = new MemoryStream())
{
var serializer = new DataContractSerializer(Value.GetType());
serializer.WriteObject(ms, Value);
info.AddValue("Bytes", ms.ToArray());
info.AddValue("IsDataContract", true);
}
}
else if (Value.GetType().IsSerializable)
{
info.AddValue("Value", Value);
info.AddValue("IsDataContract", false);
}
info.AddValue("Type", Value.GetType());
}
else
{
info.AddValue("IsNull", true);
}
}
public Wrapper(SerializationInfo info, StreamingContext context)
{
if (!info.GetBoolean("IsNull"))
{
var type = info.GetValue("Type", typeof(Type)) as Type;
if (info.GetBoolean("IsDataContract"))
{
using (var ms = new MemoryStream(info.GetValue("Bytes", typeof(byte[])) as byte[]))
{
var serializer = new DataContractSerializer(type);
Value = serializer.ReadObject(ms);
}
}
else
{
Value = info.GetValue("Value", type);
}
}
}
}
As an extension to the provided answer, I added these two methods to ease storing/retrieving the data.
public static void Set<T>(HttpSessionStateBase session, string key, T value)
{
session[key] = new Wrapper(value);
}
public static T Get<T>(HttpSessionStateBase session, string key)
{
object value = session[key];
if (value != null && typeof(T) == value.GetType())
{
return (T) value;
}
Wrapper wrapper = value as Wrapper;
return (T) ((wrapper == null) ? null : wrapper.Value);
}
This makes it a little easier to set/get values from the session:
MyDataContract c = ...;
Wrapper.Set(Session, "mykey", c);
c = Wrapper.Get<MyDataContract>(Session, "mykey");
To make it even easier, add extension methods:
public static class SessionWrapperEx
{
public static void SetWrapped<T>(this HttpSessionStateBase session, string key, T value)
{
Wrapper.Set<T>(session, key, value);
}
public static T GetWrapped<T>(this HttpSessionStateBase session, string key)
{
return Wrapper.Get<T>(session, key);
}
}
And use as below:
MyDataContract c = ...;
Session.SetWrapped("mykey", c);
c = Session.GetWrapped<MyDataContract>("mykey");

Resources