Before Playframework 2.4 I used create object for singleton class, and now I see than Singletons can be achieved using the #Singleton annotation.
But, what is the difference?
Is only for using a dependency injection with #Inject() ?
I am afraid that Playframework 2.4 with dependency Injection enter in an unnecessarily complicated way.
#Singleton is part of the JSR 330 standard and is indeed for dependency injection only. It tells your DI framework to create only one instance of the given class and use that single instance across your application. From the docs:
New instances are created every time a component is needed. If a component is used more than once, then, by default, multiple instances of the component will be created. If you only want a single instance of a component then you need to mark it as a singleton.
You can think of it as a directive to your DI framework to create only one instance. Nothing holds you back from creating multiple #Singleton annotated classes by hand using new.
Singleton objects in scala in contrast are true singletons and cannot be instantiated by hand.
Related
I have been refactoring my Play app from using Guice to using Compile-time DI.
In Guice, when we don't decorate a class with #Singleton, many instances can be created as needed.
In compile-time DI, we create an instance to be injected once, thus I think it is equivalent to a singleton.
My question is if I would lose any performance by restricting everything to be only one instance. For example, if I have an instance serviceA, with method doSomething, and considering that everything is stateless. If I have a 32-core CPU, and lots of requests come in. Would Play, in the context of compile-time DI, be able to utilize the full capacity of the CPU?
AFAiK Guice (and other runtime DI frameworks) doesn't by default produce singletons for the sole reason to be faster when creating the instances and simplify complex (potentially cyclic) dependency graph. Their goal is to start faster.
Whether you have 1 or 2 instances of ServiceA will not affect the performance of using these instances once they are created.
It's theorically even better to have singletons.
I'm using Akka Persistence with Cluster Sharding. What is the proper way to provide dependencies into such PersistentActor-s?
As far as I understand, passing them as constructor arguments is not possible, as Cluster Sharding is creating these actors.
Using Spring/Guice/etc. is not idiomatic Scala (and possibly has other issues (?)).
Using an object to implement a singleton makes for cumbersome testing and seems bad style.
What is the proper way?
P.S. If you plan to suggest the Cake pattern, please provide sample code in this specific Akka Persistence Cluster Sharding context.
UPDATED VERSION:
the solution I offered earlier did not allow to mock services of the actor under test in unit test cases.
I am using instead one solution offered on that article http://letitcrash.com/post/55958814293/akka-dependency-injection that is called "aspect weaving" and that consists of injecting the dependencies in the actor using aspect oriented programming.
This solution can be used to inject Spring dependencies on any bean not controlled by Spring container (potentially useful for legacy code).
A full example is provided by the above article: https://github.com/huntc/akka-spring/blob/f137c98b621517301f636e6ea03519388fcd5fff/src/main/scala/org/typesafe/Akkaspring.scala
And to enable aspect weaving in a spring based application you should check the documentation on Spring doc
In my case, on a jetty application server, it consists of using the spring agent and setting it in the jvm arguments.
As far as tests are concerned, I :
created setters for the injected services
created basic configuration for my actors with null beans referenced for my dependencies
instantiated the actor in my test case
replace the actor's services with mocks
run the actor's inner methods and check the results, actor's state or calls to dependencies
ORIGINAL:
I am using Akka in a Spring Application to enable clustering. At first
it raises the following issue: you cannot inject spring managed
dependencies in the actor constructor, as you said. (it tries to
serialize the application context and fails)
So I created a class that holds the application context and provides a
static method to retrieve beans I need. I retrieve the bean only if I
need it, this way:
public void onReceive{
if (message instanceof HandledMessage) {
(MyService) SpringApplicationContext.getBean("myService");
...
}
}
It's not conventional but it does the job, what do you think? Hope
otherwise it might help another one.
I'm working on my first Scala application, where we use an ActiveRecord style to retrieve data from MongoDB.
I have models like User and Category, which all have a companion object that uses the trait:
class MongoModel[T <: IdentifiableModel with CaseClass] extends ModelCompanion[T, ObjectId]
ModelCompanion is a Salat class which provide common MongoDB crud operations.
This permits to retrieve data like this:
User.profile(userId)
I never had any experience with this ActiveRecord query style. But I know Rails people are using it. And I think I saw it on Play documentation (version 1.2?) to deal with JPA.
For now it works fine, but I want to be able to run integration tests on my MongoDB.
I can run an "embedded" MongoDB with a library. The big deal is that my host/port configuration are actually kind of hardcoded on the MongoModel class which is extended by all the model companions.
I want to be able to specify a different host/port when I run integration tests (or any other "profile" I could create in the future).
I understand well dependency injection, using Spring for many years in Java, and the drawbacks of all this static stuff in my application. I saw that there is now a scala-friendly way to configure a Spring application, but I'm not sure using Spring is appropriate in Scala.
I have read some stuff about the Cake pattern and it seems to do what I want, being some kind of typesafe, compile-time-checked spring context.
Should I definitely go to the Cake pattern, or is there any other elegant alternative in Scala?
Can I keep using an ActiveRecord style or is it a total anti-pattern for testability?
Thanks
No any static references - using Cake pattern you got 2 different classes for 2 namespaces/environments, each overriding "host/port" resource on its own. Create a trait containing your resources, inherit it 2 times (by providing actual information about host/port, depending on environment) and add to appropriate companion objects (for prod and for test). Inside MongoModel add self type that is your new trait, and refactor all host/port references in MongoModel, to use that self type (your new trait).
I'd definitely go with the Cake Pattern.
You can read the following article with show an example of how to use the Cake Pattern in a Play2 application:
http://julien.richard-foy.fr/blog/2011/11/26/dependency-injection-in-scala-with-play-2-it-s-free/
The JPA 2.0 specification mentions in section 6.9 that CriteriaQuery objects are serializable, and hence may outlive any open EntityManagers or EntityManagerFactory instances:
CriteriaQuery objects must be serializable. A persistence vendor is required to support the subse- quent deserialization of a CriteriaQuery object into a separate JVM instance of that vendor’s runt- ime, where both runtime instances have access to any required vendor implementation classes.
The EJB 3.1 specification says in section 21.2.2:
An enterprise bean must not use thread synchronization primitives to synchronize execution of multiple instances, except if it is a Singleton session bean with bean-managed concurrency.
If I have a stateless session bean that wishes to pre-build a bunch of CriteriaQuery objects using a CriteriaBuilder obtained from an injected #PersistenceContext, where should I stash the results?
I can think of the following possibilities but am concerned that all but one run afoul of the "no synchronization primitives" clause above:
In a Map that is stored as the value of one of my bean's instance fields, understanding that I'll have to synchronize access to the map. My take: section 21.2.2 violation.
In a ConcurrentMap that is stored as the value of one of my bean's instance fields. My take: still a section 21.2.2 violation, as I'm sure the ConcurrentMap implementation synchronizes somewhere.
In a #Singleton EJB's instance field somewhere, where the #Singleton exists only to serve as this kind of cache; with bean-managed concurrency this should be legal, but now all my stateless session beans that want to make use of this CriteriaQuery cache have to inject the singleton into themselves...seems like a lot of overhead.
So it sounds like strictly speaking the last option is the only specification-compliant one. Am I correct?
I would consider putting them in a simple static context, accessible from anywhere. The problem lies in initializing them since you need an entity manager instance to do that. Perhaps a singleton ejb for initializing things as described at Call method in EJB on JBoss startup. The singleton could initialize your criteria query cache, which then could serve criteria queries to your DAOs through static context.
Another option would be to use JPQL which has built in support for precompiled queries. Of course you'd lose some advantages of using the criteria API, though I think the main issue (type safety etc.) might be OK since precompiled queries should throw an exception if they are invalid at deploy time rather than runtime.
I have set up my project to work with Entity Framework. It is a database first approach.
For the IoC, I am using Castle Windsor and the project is set up to inject all the dependencies.
My question is, is there a need to set up Castle to do the same for my EF entity? How can I do that?
There are a number of ways to use Entity Framework with Castle. I would suggest that you adopt both the Unit of Work and Repository patterns. The DbContext provides most of the functionality for a unit of work, and DbSet similarly provides most of what is needed for a repository.
The main issue you will run into is that EF doesn't understand dependency injection for model objects. EF always uses the empty constructor when it allocates a model object, and this means we don't get the opportunity to do dependency injection into model objects. We have gotten around this by using the object materialization hook on the DbContext:
((IObjectContextAdapter)this).ObjectContext.ObjectMaterialized += EFObjectLoadInitializer;
The EFObjectLoadInitializer then injects the Kernel into the model object (all our model objects must support the IInjectKernel interface). Upon injection, the Kernel is then used to resolve any other dependencies. Its not very clean, but it works well and allows us to leverage off of the rest of the infrastructure that we have set up with Windsor.