I'm trying to fully decorate an API that uses (what I call) the throwOnError pattern.
I know some people dislike it but this is not the point: it is used by people far more clever than me, for instance in the Assembly.GetType method:
public virtual Type? GetType(string name, bool throwOnError)
I'm a little bit disappointed: it seems that currently (C#9/net5), it is not possible to state that the return cannot be null as long as throwOnError is true. Something like: [return: NotNullIfTrue( "throwOnError" )].
The source here is (sadly) naked: https://source.dot.net/#System.Private.CoreLib/Assembly.cs,106
Does anyone has a solution to this? Should we wait for a NotNullIfTrue attribute?
PS: I don't want to use [DoesNotReturnIf(true)] since this is obvisoulsy a too strong assertion.
Related
I just started using the Netbeans 7.1 beta and it is calling out errors of a type which I have never seen before. Specifically:
A managed bean with a public field should not declare any scope other than #Dependent.
The fields it is complaining about are public static final. I can understand the restriction on non-static fields, but I can't think of a good reason this would not be allowed for a static field. Unfortunately I use a lot of them since I don't like having constants in my code.
I note that even though I get the red dot in the margin in the editor, the maven-driven build still works and GlassFish still runs my application the way I would expect.
So what is my denoument on this issue? Am I going to have to move my static fields elsewhere or is there another way of handling this?
Quoting the javax.enterprise.inject package javadocs:
If a managed bean has a public field, it must have scope #Dependent.
But I do agree wih #BalusC that if this compiles, Netbeans should report it as Warning (does it?).
Anyway, are those constants really part of the API? I mean, do you access they anywhere else but within their own classes? If not, reduce visibility to private. (If you just need to access the constants from the view you can also create accessors for the private constant). If yes, I would suggest you to move them somewhere else anyway.
Public fields (static or not) aren't proxyable - that's why they can only be dependent scoped. To work around this you obviously can access them through getter methods.
Say I have a class that looks like the following:
internal class SomeClass
{
IDependency _someDependency;
...
internal string SomeFunctionality_MakesUseofIDependency()
{
...
}
}
And then I want to add functionality that is related but makes use of a different dependency to achieve its purpose. Perhaps something like the following:
internal class SomeClass
{
IDependency _someDependency;
IDependency2 _someDependency2;
...
internal string SomeFunctionality_MakesUseofIDependency()
{
...
}
internal string OtherFunctionality_MakesUseOfIDependency2()
{
...
}
}
When I write unit tests for this new functionality (or update the unit tests that I have for the existing functionality), I find myself creating a new instance of SomeClass (the SUT) whilst passing in null for the dependency that I don't need for the particular bit of functionality that I'm looking to test.
This seems like a bad smell to me but the very reason why I find myself going down this path is because I found myself creating new classes for each piece of new functionality that I was introducing. This seemed like a bad thing as well and so I started attempting to group similar functionality together.
My question: should all dependencies of a class be consumed by all its functionality i.e. if different bits of functionality use different dependencies, it is a clue that these should probably live in separate classes?
When every instance method touches every instance variable then the class is maximally cohesive. When no instance method shares an instance variable with any other, the class is minimally cohesive. While it is true that we like cohesion to be high, it's also true that the 80-20 rule applies. Getting that last little increase in cohesion may require a mamoth effort.
In general if you have methods that don't use some variables, it is a smell. But a small odor is not sufficient to completely refactor the class. It's something to be concerned about, and to keep an eye on, but I don't recommend immediate action.
Does SomeClass maintain an internal state, or is it just "assembling" various pieces of functionality? Can you rewrite it that way:
internal class SomeClass
{
...
internal string SomeFunctionality(IDependency _someDependency)
{
...
}
internal string OtherFunctionality(IDependency2 _someDependency2)
{
...
}
}
In this case, you may not break SRP if SomeFunctionality and OtherFunctionality are somehow (functionally) related which is not apparent using placeholders.
And you have the added value of being able to select the dependency to use from the client, not at creation/DI time. Maybe some tests defining use cases for those methods would help clarifying the situation: If you can write a meaningful test case where both methods are called on same object, then you don't break SRP.
As for the Facade pattern, I have seen it too many times gone wild to like it, you know, when you end up with a 50+ methods class... The question is: Why do you need it? For efficiency reasons à la old-timer EJB?
I usually group methods into classes if they use a shared piece of state that can be encapsulated in the class. Having dependencies that aren't used by all methods in a class can be a code smell but not a very strong one. I usually only split up methods from classes when the class gets too big, the class has too many dependencies or the methods don't have shared state.
My question: should all dependencies of a class be consumed by all its functionality i.e. if different bits of functionality use different dependencies, it is a clue that these should probably live in separate classes?
It is a hint, indicating that your class may be a little incoherent ("doing more than just one thing"), but like you say, if you take this too far, you end up with a new class for every piece of new functionality. So you would want to introduce facade objects to pull them together again (it seems that a facade object is exactly the opposite of this particular design rule).
You have to find a good balance that works for you (and the rest of your team).
Looks like overloading to me.
You're trying to do something and there's two ways to do it, one way or another. At the SomeClass level, I'd have one dependency to do the work, then have that single dependent class support the two (or more) ways to do the same thing, most likely with mutually exclusive input parameters.
In other words, I'd have the same code you have for SomeClass, but define it as SomeWork instead, and not include any other unrelated code.
HTH
A Facade is used when you want to hide complexity (like an interface to a legacy system) or you want to consolidate functionality while being backwards compatible from an interface perspective.
The key in your case is why you have the two different methods in the same class. Is the intent to have a class which groups together similar types of behavior even if it is implemented through unrelated code, as in aggregation. Or, are you attempting to support the same behavior but have alternative implementations depending on the specifics, which would be a hint for a inheritance/overloading type of solution.
The problem will be whether this class will continue to grow and in what direction. Two methods won't make a difference but if this repeats with more than 3, you will need to decide whether you want to declare it as a facade/adapter or that you need to create child classes for the variations.
Your suspicions are correct but the smell is just the wisp of smoke from a burning ember. You need to keep an eye on it in case it flares up and then you need to make a decision as how you want to quench the fire before it burns out of control.
I have a question, and I'm going to tag this subjective since that's what I think it evolves into, more of a discussion. I'm hoping for some good ideas or some thought-provokers. I apologize for the long-winded question but you need to know the context.
The question is basically:
How do you deal with concrete types in relation to IoC containers? Specifically, who is responsible for disposing them, if they require disposal, and how does that knowledge get propagated out to the calling code?
Do you require them to be IDisposable? If not, is that code future-proof, or is the rule that you cannot use disposable objects? If you enforce IDisposable-requirements on interfaces and concrete types to be future-proof, whose responsibility is objects injected as part of constructor calls?
Edit: I accepted the answer by #Chris Ballard since it's the closest one to the approach we ended up with.
Basically, we always return a type that looks like this:
public interface IService<T> : IDisposable
where T: class
{
T Instance { get; }
Boolean Success { get; }
String FailureMessage { get; } // in case Success=false
}
We then return an object implementing this interface back from both .Resolve and .TryResolve, so that what we get in the calling code is always the same type.
Now, the object implementing this interface, IService<T> is IDisposable, and should always be disposed of. It's not up to the programmer that resolves a service to decide whether the IService<T> object should be disposed or not.
However, and this is the crucial part, whether the service instance should be disposed or not, that knowledge is baked into the object implementing IService<T>, so if it's a factory-scoped service (ie. each call to Resolve ends up with a new service instance), then the service instance will be disposed when the IService<T> object is disposed.
This also made it possible to support other special scopes, like pooling. We can now say that we want minimum 2 service instances, maximum 15, and typically 5, which means that each call to .Resolve will either retrieve a service instance from a pool of available objects, or construct a new one. And then, when the IService<T> object that holds the pooled service is disposed of, the service instance is released back into its pool.
Sure, this made all code look like this:
using (var service = ServiceContainer.Global.Resolve<ISomeService>())
{
service.Instance.DoSomething();
}
but it's a clean approach, and it has the same syntax regardless of the type of service or concrete object in use, so we chose that as an acceptable solution.
Original question follows, for posterity
Long-winded question comes here:
We have a IoC container that we use, and recently we discovered what amounts to a problem.
In non-IoC code, when we wanted to use, say, a file, we used a class like this:
using (Stream stream = new FileStream(...))
{
...
}
There was no question as to whether this class was something that held a limited resource or not, since we knew that files had to be closed, and the class itself implemented IDisposable. The rule is simply that every class we construct an object of, that implements IDisposable, has to be disposed of. No questions asked. It's not up to the user of this class to decide if calling Dispose is optional or not.
Ok, so on to the first step towards the IoC container. Let's assume we don't want the code to talk directly to the file, but instead go through one layer of indirection. Let's call this class a BinaryDataProvider for this example. Internally, the class is using a stream, which is still a disposable object, so the above code would be changed to:
using (BinaryDataProvider provider = new BinaryDataProvider(...))
{
...
}
This doesn't change much. The knowledge that the class implements IDisposable is still here, no questions asked, we need to call Dispose.
But, let's assume that we have classes that provide data that right now doesn't use any such limited resources.
The above code could then be written as:
BinaryDataProvider provider = new BinaryDataProvider();
...
OK, so far so good, but here comes the meat of the question. Let's assume we want to use an IoC container to inject this provider instead of depending on a specific concrete type.
The code would then be:
IBinaryDataProvider provider =
ServiceContainer.Global.Resolve<IBinaryDataProvider>();
...
Note that I assume there is an independent interface available that we can access the object through.
With the above change, what if we later on want to use an object that really should be disposed of? None of the existing code that resolves that interface is written to dispose of the object, so what now?
The way we see it, we have to pick one solution:
Implement runtime checking that checks that if a concrete type that is being registered implements IDisposable, require that the interface it is exposed through also implements IDisposable. This is not a good solution
Enfore a constraint on the interfaces being used, they must always inherit from IDisposable, in order to be future-proof
Enforce runtime that no concrete types can be IDisposable, since this is specifically not handled by the code using the IoC container
Just leave it up to the programmer to check if the object implements IDisposable and "do the right thing"?
Are there others?
Also, what about injecting objects in constructors? Our container, and some of the other containers we've looked into, is capable of injecting a fresh object into a parameter to a constructor of a concrete type. For instance, if our BinaryDataProvider need an object that implements the ILogging interface, if we enforce IDispose-"ability" on these objects, whose responsibility is it to dispose of the logging object?
What do you think? I want opinions, good and bad.
One option might be to go with a factory pattern, so that the objects created directly by the IoC container never need to be disposed themselves, eg
IBinaryDataProviderFactory factory =
ServiceContainer.Global.Resolve<IBinaryDataProviderFactory>();
using(IBinaryDataProvider provider = factory.CreateProvider())
{
...
}
Downside is added complexity, but it does mean that the container never creates anything which the developer is supposed to dispose of - it is always explicit code which does this.
If you really want to make it obvious, the factory method could be named something like CreateDisposableProvider().
(Disclaimer: I'm answering this based on java stuff. Although I program C# I haven't proxied anything in C# but I know it's possible. Sorry about the java terminology)
You could let the IoC framework inspect the object being constructed to see if it supports
IDisposable. If not, you could use a dynamic proxy to wrap the actual object that the IoC framework provides to the client code. This dynamic proxy could implement IDisposable, so that you'd always deliver a IDisposable to the client. As long as you're working with interfaces that should be fairly simple ?
Then you'd just have the problem of communicating to the developer when the object is an IDisposable. I'm not really sure how this'd be done in a nice manner.
You actually came up with a very dirty solution: your IService contract violates the SRP, wich is a big no-no.
What I recommend is to distinguish so-called "singleton" services from so-called "prototype" services. Lifetime of "singleton" ones is managed by the container, which may query at runtime whether a particular instance implements IDisposable and invoke Dispose() on shutdown if so.
Managing prototypes, on the other hand, is totally the responsibility of the calling code.
I want to handle different types of docs the same way in my application
Therefore:
I have a generic interface like this.
public interface IDocHandler<T>where T: class
{
T Document { get;set;}
void Load(T doc);
void Load(string PathToDoc);
void Execute();
void Execute(T doc);
}
And for different types of documents I implement this interface.
for example:
public class FinanceDocumentProcessor:IDocumentHandler<ReportDocument>
{}
public class MarketingDocumentProcessor:IDocumentHandler<MediaDocument>
{}
Then I can do of course something like this:
IDocumentHandler<ReportDocument> docProc= new FinanceDocumentProcessor();
It would be interessting to know how I could inject T at runtime to make the line above loosly coupled...
IDocumentHandler<ReportDocument> docProc = container.resolve("FinanceDocumentProcessor());
but I want to decide per Configuration wether I want to have my FinanceDomcumentProcessor or my MarketingDocumentProcessor... therefore I would have to inject T on the left site, too ...
Since I have to use c# 2.0 I can not use the magic word "var" which would help a lot in this case... but how can I design this to be open and flexible...
Sorry for the misunderstanding and thanks for all the comments but I have another example for my challenge (maybe I am using the wrong design for that) ...
But I give it a try: Same situation but different Explanation
Example Image I have:
ReportingService, Crystal, ListAndLabel
Three different Reporting Document types. I have a generic Handler IReportHandler<T> (would be the same as above) this Handler provides all the functionality for handling a report Document.
for Example
ChrystalReportHandler:IReportHandler<CrystalReportDocument>
Now I want to use a Framework like Unity (or some else framework) for dependency injection to decide via configuration whether I want to use Crystal, Reportingservices or List and Label.
When I specify my mapping I can inject my ChrystalReportHandler but how can I inject T on the left side or in better word The Type of ReportDocument.
IReportHandler<T (this needs also to be injected)> = IOContainer.Resolve(MyMappedType here)
my Problem is the left Site of course because it is coupled to the type but I have my mapping ... would it be possible to generate a object based on Mapping and assign the mapped type ? or basically inject T on the left side, too?
Or is this approach not suitable for this situation.
I think that with your current design, you are creating a "dependency" between IDocumentHandler and a specific Document (ReportDocument or MediaDocument) and so if you want to use IDocumentHandler<ReportDocument or MediaDocument> directly in your code you must assume that your container will give you just that. The container shouldn't be responsible for resolving the document type in this case.
Would you consider changing your design like this?
public interface IDocumentHandler
{
IDocument Document { get; set; }
void Load(IDocument doc);
void Load(string PathToDoc);
void Execute();
void Execute(IDocument doc);
}
public class IDocument { }
public class ReportDocument : IDocument { }
public class MediaDocument : IDocument { }
public class FinanceDocumentProcessor : IDocumentHandler { }
public class MarketingDocumentProcessor : IDocumentHandler { }
If I understand you correctly, you have two options.
if you have interface IDocHandler and multiple classes implementing it, you have to register each type explicitly, like this:
container.AddComponent>(typeof(FooHandler));
if you have one class DocHandler you can register with component using open generic type
container.AddComponent(typeof(IDocHandler<>), typeof(DocHandler<>));
then each time you resolve IDocHandler you will get an instance of DocHandler and when you resolve IDocHandler you'll get DocHandler
hope that helps
You need to use a non-generic interface on the left side.
Try:
public interface IDocumentHandler { }
public interface IDocumentHandler<T> : IDocumentHandler { }
This will create two interfaces. Put everything common, non-T-specific into the base interface, and everything else in the generic one.
Since the code that you want to resolve an object into, that you don't know the type of processor for, you couldn't call any of the T-specific code there anyway, so you wouldn't lose anything by using the non-generic interface.
Edit: I notice my answer has been downvoted. It would be nice if people downvoting things would leave a comment why they did so. I don't care about the reputation point, that's just minor noise at this point, but if there is something seriously wrong with the answer, then I'd like to know so that I can either delete the answer (if it's way off target) or correct it.
Now in this case I suspect that either the original questionee has downvoted it, and thus either haven't posted enough information, so that he's actually asking about something other than what he's asked about, or he didn't quite understand my answer, which is understandable since it was a bit short, or that someone who didn't understand it downvoted it, again for the same reason.
Now, to elaborate.
You can't inject anything "on the left side". That's not possible. That code have to compile, be correct, and be 100% "there" at compile-time. You can't say "we'll tell you what T is at runtime" for that part. It just isn't possible.
So the only thing you're left with is to remove the T altogether. Make the code that uses the dependency not depend on T, at all. Or, at the very least, use reflection to discover what T is and do things based on that knowledge.
That's all you can do. You can't make the code on the left side change itself depending on what you return from a method on the right side.
It isn't possible.
Hence my answer.
What is the best way of adding a new object in the entity framework. The designer adds all these create methods, but to me it makes more sense to call new on an object. The generated CreateCustomer method e.g. could be called like this:
Customer c = context.CreateCustomer(System.Guid.NewGuid(), "Name"));
context.AddToCustomer(c);
where to me it would make more sense to do:
Customer c = new Customer {
Id = System.Guid.NewGuid(),
Name = "Name"
};
context.AddToCustomer(c);
The latter is much more explicit since the properties that are being set at construction are named. I assume that the designer adds the create methods on purpose. Why should I use those?
As Andrew says (up-voted), it's quite acceptable to use regular constructors. As for why the "Create" methods exist, I believe the intention is to make explicit which properties are required. If you use such methods, you can be assured that you have not forgotten to set any property which will throw an exception when you SaveChanges. However, the code generator for the Entity Framework doesn't quite get this right; it includes server-generated auto increment properties, as well. These are technically "required", but you don't need to specify them.
You can absolutely use the second, more natural way. I'm not even sure of why the first way exists at all.
I guess it has to do with many things. It looks like factory method to me, therefore allowing one point of extension. 2ndly having all this in your constructor is not really best practice, especially when doing a lot of stuff at initialisation. Yes, your question seems reasonable, i even agree with it, however, in terms of object design, it is more practical as they did it.
Regards,
Marius C. (c_marius#msn.com)