I've seen a class declared with its only constructor being annotated with #Inject.
And I don't see the one constructor being called anywhere in the entire project.
So two questions:
<1> What does #Inject mean? (What does it do? Why is the constructor being annotated with it?)
<2> As mentioned, the constructor never gets called directly, does that have anything to do with the fact that it is annotated with #Inject?
Google Guice is a dependency injection library that allows you to construct objects simply by declaring relationships between them. Objects are constructed as they are demanded to construct other objects. You can also implement abstract classes or interfaces with different implementations by configuring Guice, which makes it very useful for running or testing your code.
#Inject annotates constructors and methods that determine what an object needs to be initialized. There are also a lot of other annotations that determine how Guice works. But simply annotating objects isn't enough; you also have to configure them with Guice bindings.
Here's a really simple example (from one of my applications). I have a MySQLDataTracker that requires a MysqlConnectionPoolDataSource:
public class MySQLDataTracker extends ExperimentDataTracker {
#Inject
public MySQLDataTracker(MysqlConnectionPoolDataSource ds) {
....
}
}
Note that MySQLDataTracker extends ExperimentDataTracker, an abstract class that can be implemented several ways. In my Guice bindings I declare that
bind(ExperimentDataTracker.class).to(MySQLDataTracker.class);
This declares that whenever I want an ExperimentDataTracker, a MySQLDataTracker will be constructed. I also need to make sure that the requisite object for constructing this is available, so I declare a provider:
#Provides #Singleton
MysqlConnectionPoolDataSource getMysqlCPDS() {
return (some thingy I construct...);
}
This says that there should only be a single connection pool data source. It also means that when I try to get an instance of ExperimentDataTracker, Guice has everything it needs to construct it. If I didn't have the above, it would throw an error.
ExperimentDataTracker tracker = injector.getInstance(ExperimentDataTracker.class);
However, it doesn't stop here. Other things depend on the ExperimentDataTracker, so it's used in turn to inject other objects. At the top level of my code there is actually only one call to getInstance, which makes Guice construct pretty much everything. I don't have to write the new statement anywhere.
I'm a big fan of Guice after seeing how it reduced the need for me to initialize a bunch of objects in order to initialize other objects. Basically I just ask for the object I want, and poof! it appears.
Related
I'm a fairly experienced programmer but I'm fairly new to Guice and I'm not sure what the best practices are. Here is my possibly flawed understanding so far, please point out if I use incorrect terminology, question follows after:
Classes with a (public) no arguments constructor or an #Inject-annotated constructor (there may be only one such constructor) do not need to be explicitly bound/provided in the module. Though in the latter case, of course, the constructor parameters may need to be explicitly provided unless they can also be auto-provided by the same rule.
When instances of such classes are (automatically) provided, this can always be thought of as being done in injector.getInstance(...) fashion – #Inject-annotated members (like say injected private fields in the class, see my Car example) will also be injected and not left null.
The above two facts(?) partially solve the problem that one can not explicitly bind/provide an instance with #Inject-annotated members, since one can not create an instance inside a provider using injector.getInstance(...) and so the #Inject-annotated members will be left null.
The remaining problem case is when the constructor requires one or more parameters to be given during creation, the values of which cannot be known up front and therefore cannot be injected.
The solution generally seems to be to provide/inject a factory with a parameterized create method to create the instances when and as needed, but this could only be done using new, which does not work in general – what if the class contains #Inject-annotated members itself? Constructing the class with new will leave those should-be-injected members null-valued.
Here is sample code:
class Car {
#Inject
private Engine engine;
...
public Car(final String cannotBeInjected) {
...
}
}
How would I make Car (or a Car factory) injectable and still allow for an Engine instance to also be injected? A possible solution is to just assign the cannotBeInjected value via some setter after creation, but this just doesn't seem elegant.
Hi I have a very simple dagger questions for android.
class Fooz {
#Inject Foo1 mFoo1;
public Fooz() {
....
}
}
class Fooz {
private Foo1 mFoo1;
#Inject public Fooz(Foo1 foo1) {
mFoo1 = foo1;
}
}
How are the two classes identical?
The first one injects Foo1 field directly while the second one assignes mFoo1 in the constructor.
For the second one, does Foo1 get injected from object graph as soon as Fooz is created and added to object graph?
If they are different, why so?
Thanks!
Constructor injection gives you more control over the object instantiation since using field injections means to restrict your class creation to reflection and rely on support to these particular injection annotations. Besides that, having the dependencies clearly on the constructor let the code easier to maintain and to test.
As far as I know, there is no difference regarding the way it is held on the dagger graph but a constructor call is always faster than injected fields.
In my opinion, we should use property only when we do not have control over the object creation, as in Activities and Fragments, per example.
These classes will behave the same when Fooz will be Injected using dependency injection. However they will behave differently when constructed using Constructor's you defined.
Example 1. Calling new Fooz() will result in mFoo1 being null.
Example 2. Calling new Fooz(foo1) will result in mFoo1 being initialized to foo1.
The preferred (personal opinion) way is to use dependency injection annotation on constructor, because it will avoid null pointer exceptions, as explained when comparing example 1 and example 2. What is more such constructor gives more flexibility when testing your classes as you can provide mocks, much easier.
These is sonarqube rule with better description, explaining what I mentioned https://sonarcloud.io/coding_rules?open=squid%3AS3306&rule_key=squid%3AS3306 .
If I follow dependency injection principle for a class design, I should make sure my class does not try to instantiate its dependencies within my class, but rather ask for the object through the constructor.
This way I am in control of the dependencies I provide for the class while unit-testing it. This I understand.
But what I am not sure is does this mean that a good class design which follows dependency injection principle means that its fields should never be initialized inline? Should we totally avoid inline initialization to produce testable code?
EDIT
Which is better 1 or 2?
1
public class Car {
private Tire tire = new Tire(); //
}
2
public class Car {
private Tire tire;
public Car(Tire tire) {
this.tire = tire
}
}
No, it surely doesn't mean inline initialization.
I'm not a Java user but this term is relevant to many programming languages.
Basically, instead of having your objects creating a dependency or asking a factory object to make one for them, you pass the needed dependencies into the object externally, and you make it somebody else's problem.
public SomeClass() {
myObject = Factory.getObject();
}
"Dependency Injection" is a 25-dollar term for a 5-cent concept. [...]
Dependency injection means giving an object its instance variables.
[...].
Dependency injection is basically providing the objects that an object needs (its dependencies) instead of having it construct them itself. It's a very useful technique for testing since it allows dependencies to be mocked or stubbed out.
Any application is composed of many objects that collaborate with each other to perform some useful stuff. Traditionally each object is responsible for obtaining its own references to the dependent objects (dependencies) it collaborates with. This leads to highly coupled classes and hard-to-test code.
A lot of reference from here
In my opinion, In general, if you think that the data field is a dependency, then let the dependency management container should manage it. What counts a dependency and what is not dependency is a tough question, and only you can decide.
For example: if your design of car allows working with different types of Tyres, then it's a dependency. Otherwise, inside a class Car you'll have to work with some kind of TyreFactory, which is not really reasonable.
If you work with DI container, testing the class will be a breeze. You'll have to provide a stub/mock to the class which is obviously a real benefit. So, in your example, if you go with the first solution then, how will you test that your car works with different types of Tyres?
In some places where a class hierarchy is present and the top most base class is an abstract class there is a static getInstance() method in the abstract class. This will be responsible for creating the correct sub-class and returning it to the caller. For example consider the below code.
public class abstract Product {
public static Product getInstance(String aCode) {
if ("a".equals(aCode) {
return new ProductA();
}
return ProductDefault();
}
// product behaviour methods
}
public class ProductA extends Product {}
public class ProductDefault extends Product {}
In Java, java.util.Calendar.getInstance() is one place this pattern has been followed. However this means each time a new subclass is introduced one has to modify the base class. i.e: Product class has to be modified in the above example. This seems to violate the ocp principle. Also the base class is aware about the sub class details which is again questionable.
My question is...
is the above pattern an anti-pattern ?
what are the draw-backs of using the above pattern ?
what alternatives can be followed instead ?
The interface is not an anti-pattern. But the way you've implemented it is rather poor ... for the reason you identified. A better idea would be to have some mechanism for registering factory objects for each code:
The Java class libraries do this kind of thing using SPIs and code that looks reflectively for "provider" classes to be dynamically loaded.
A simpler approach is to have a "registry" object, and populate it using dependency injection, or static initializers in the factory object classes, or a startup method that reads class names from a properties file, etcetera.
No it's not. It's more like factory method pattern http://en.wikipedia.org/wiki/Factory_method_pattern. E.g. Calendar.getInstance();. JDK is full of such examples. Also reminds of Effective Java Item 1: Consider static factory methods instead of constructors
There are a number of separate issues here.
getInstance is probably going to be a bad name. You explicitly want a new object you can play around with. "Create", "make", "new" or just leave that word out. "Instance" is also a pretty vacuous word in this context. If there is sufficient context from the class name leave it out, otherwise say what it is even if that is just a type name. If the method returns an immutable object, of is the convention (valueOf in olden times).
Putting it in an abstract base class (or in an interface if that were possible) is, as identified, not the best idea. In some cases an enumeration of all possible subtypes is appropriate - an enum obviously and really not that bad if you are going to use visitors anyway. Better to put it in a new file.
Anything to do with mutable statics is wrong. Whether it is reusing the same mutable instance, registration or doing something disgusting with the current thread. Don't do it or depend (direct or indirectly) on anything that does.
Based on the feedback i introduced a new ProductFactory class that took care of creating the correct Product. In my case the creation of the correct product instance depends on an external context (i've put the product code for the purpose of simplicity.. in the actual case it might be based on several parameters.. these could change over time). So having a Product.getInstance() method is not that suited because of the reasons outlined in the question. Also having a different ProductFactory means in the future.. Product class can become an interface if required. It just gives more extensibility.
I think when the creation of the object doesn't depend on an external context.. like in the case of Calendar.getInstance() it's perfectly ok to have such a method. In these situations the logic of finding the correct instance is internal to that particular module/class and doesn't depend on any externally provided information..
I want to intercept all method invocations to some class MyClass to be able to react on some setter-invocations.
I tried to use dynamic proxies, but as far as I know, this only works for classes implementing some interface. But MyClass does not have such an interface.
Is there any other way, besides implementing a wrapper class, that delegates all invocations to a member, which is an instance of the MyClass or besided using AOP?
As you note, you cannot use JDK dynamic proxies (no interface), but using Spring and CGLIB (JAR included with Spring), you can do the following:
public class Foo
{
public void setBar()
{
throw new UnsupportedOperationException("should not go here");
}
public void redirected()
{
System.out.println("Yiha");
}
}
Foo foo = new Foo();
ProxyFactory pf = new ProxyFactory(foo);
pf.addAdvice(new MethodInterceptor()
{
public Object invoke(MethodInvocation mi) throws Throwable
{
if (mi.getMethod().getName().startsWith("set"))
{
Method redirect = mi.getThis().getClass().getMethod("redirected");
redirect.invoke(mi.getThis());
}
return null;
}
});
Foo proxy = (Foo) pf.getProxy();
proxy.setBar(); // prints "Yiha"
If you are prepared to do something really ugly, have a look at:
http://docs.oracle.com/javase/7/docs/technotes/guides/jpda/
Basically the debugger interface ought to allow you to attach like a debugger, and hence intercept calls. Bear in mind I think this is a really bad idea, but you asked if it was possible.
Java doesn't have any actual language features for method interception (not sure any static language does)
I kinda like Nick's idea of using the debugger interface, that's just mean.
I think the short answer you need is: No there isn't a way of intercepting a method call in Java without actually replacing the class using a proxy or wrapper.
Note: The AOP libraries just make this happen automatically.
Some of the Java gurus might frown upon this but I've had some good success with avoiding primitive types and setters altogether. My class looks like this:
class Employee extends SmartPojo {
public SmartString name;
public SmartInt age;
}
You'll notice two things: 1. everything is public. 2. No constructor.
The magic happens in SmartPojo which searches for any field which implements the "Smart" interface and initializes it. Since this is no primitive (and no final class), I can add set() and get() methods for all fields anywhere in my model in a single place. So no setter/getter wastes anymore, it's stunningly simple to add notification (also in a single place), etc.
True, this is no POJO anymore and it's not a Bean in most ways but I've found that these old ideas limit me more than they help. YMMV.
I just developed a small framework for this purpose.
You can check it out at: http://code.google.com/p/java-interceptor/ (use svn to check out).
There isn't a lot of magic in AspectJ. You can write your own agent. http://java.sun.com/javase/6/docs/api/java/lang/instrument/package-summary.html seems to be good starting point.
Why cannot your class implement an interface? You could just extract some interface from it containing all the methods that you want to intercept and use the dynamic proxies mechanism easily. It's also a good programming practice to code with interfaces and not classes.
You could use Spring framework with Spring AOP capabilities (which are using dynamic proxies inside) to do it. You will just have to define your class as a Spring bean in the configuration file and clients of your class will have to either get its instance from the Spring application context or as a dependency automatically (by defining the setMyClass(MyClass mc) method for instance). From there you can easily go to defining an aspect that intercepts all the method calls to this class.