Spring Transaction with AspectJ - spring

I'm having a problem with #Transaction in Spring.
Basically, no transaction is created with the following message:
delaying identity-insert due to no transaction in progress
I posetd a full description of the problen in the Spring AOP forum
http://forum.springsource.org/showthread.php?132612-Transaction-management
Any suggestion would be appreciated!
Stefano

in your code you have defined the service like this:
#Service
#Configurable
public class ServiceImpl<T> implements Service<T> {
#Override
#Transactional
public T save(T entity) {
....
}
}
I do not know why you use the #Configurable annotation. If you do not need them (that is if you do not create an instance of this service via new) , I would remove them.
The reason is that I remember that spring docu say that #Configurable enable injection, but it does not say anything about load time weaving support enabled by #Configurable

Found the solution.
The problem was in the configuratoin. I added two elements context:spring-configured and context:load-time-weaver that mean two different things. The first one activate AspectJ compile time weaving (which I didn't understand), the second one activate AspectJ load time weaving. I think that these two elements don't work well together.
I decided to use compile time weaver and (after struggling a little bit...) I found that (obviously) external jars are not weaved with this configuration, simply because they have yet been built.
The easiest way to solve is to modify the configuration of the aspectj-maven-plugin (in POM.xml) adding under the configuration node:
<weaveDependencies>
<weaveDependency>
<groupId>GROUP-ID</groupId>
<artifactId>ARTIFACT-ID</artifactId>
</weaveDependency>
</weaveDependencies>
for every artifact you need to weave.
Now transactions work fine!
#Ralph: many thanks for your time!
Regards,
Stefano

Related

Spring #Transactional - synchronize via AspectJ

I am trying to synchronize declarative transactions (i.e. methods annotated with #Transactional) using AspectJ like so:
...
import org.aspectj.lang.annotation.Aspect;
...
#Component
#Aspect
public class TransactionMonitor extends TransactionSynchronizationAdapter {
#Before("execution(#org.springframework.transaction.annotation.Transactional * *.*(..))")
private void registerTransactionSynchronizationOnAnnotation(JoinPoint joinPoint) {
TransactionSynchronizationManager.registerSynchronization(this);
}
}
This currently fails with java.lang.IllegalStateException: Transaction synchronization is not active which indicates that the synchronization is not run inside the transaction execution, but before. I want to ensure that this is the other way round, of course.
I found this answer, however #Order(Ordered.LOWEST_PRECEDENCE) had no effect, and
#DeclarePrecedence(
"org.springframework.transaction.aspectj.AnnotationTransactionAspect, xxx.xxx.TransactionMonitor, *"
)
led to this during startup:
java.lang.IllegalArgumentException: DeclarePrecedence not presently supported in Spring AOP
I have the feeling this is AOP and AspectJ not being happy with each other, but I am not sure. I am thankful for any ideas.
EDIT: I have to use #EnableTransactionManagement(proxyTargetClass = true), can this be related to the issue?
For #DeclarePrecedence you need to switch to native AspectJ. Spring AOP is just "AOP lite" and technologically has little in common with AspectJ other than its syntax which is basically an AspectJ subset. The Spring manual describes how to use native AspectJ in Spring via LTW (load-time weaving). Precedence declaration for Spring components rather works using #Order, BTW.
I am not a Spring user at all, but as for declarative transaction management, it already knows proxy-based Spring AOP versus native AspectJ mode, see EnableTransactionManagement.mode and the enum constants in AdviceMode. Besides, EnableTransactionManagement also has an order property. Reading Javadoc and the Spring manual helps, I guess.

How #Aspect with #Component annotation works under the hood

I've been looking for an answer for a while, but no luck so far, thus I'm coming here for some words of wisdom.
I've created an aspect using #Aspect annotation, because I need to #Autowire some singleton dependencies I've decided to annotate this aspect class with #Component and let the Spring to do the magic. It works, however ...
I'm fairly familiar with AOP concept, what's weaving and different flavors of it (cglib vs aspectj) but it's not fully intuitive to me how it works under the hood.
#Component means a given class will be a singleton within a given context, #Aspect means that the content of an aspect class will be somehow weaved into the target class during runtime/compilation - and this target class is not a singleton but prototype for instance. So what I'm ending up with at the end?
Spring AOP does not do compile-time-weaving and does not modify the code of the advised target. Instead it works with proxies that are weaved around the joinpoints. That is why Spring AOP aspects and be used as (singleton) components, have their fields autowired, etc., like any other Spring Proxy.
It is also the reason why Spring AOP aspects only work for public method executions, not field accesses and the like.
The documentation is quite well written and goes into as much (or as little) detail as one might like:
http://docs.spring.io/spring/docs/current/spring-framework-reference/html/aop.html
The book AspectJ in Action's section 2.5 is on the internal working of the weaving step, it is only 2 page but gets the point across well.
Luckily the section is available here.
This is for posterity.

Is it possible to autowire dependencies that are created outside of an ApplicationContext?

I've got an application which uses JAXRS to map Restlet resources using annotations. However, the only entry point I have is essentially defining a list of resource classes in the application configuration. These classes are instantiated by Restlet or JAXRS, so I have no way to put them in my ApplicationContext. Is there a way to have Spring scan the classpath and autowire new instances as necessary? I've already tried using something like below:
#Autowired
private SessionFactory sessionFactory;
Unfortunately, it doesn't really work. Is there a way to do what I'm talking about here?
You can use AspectJ to dependency inject your beans that are created out of your control, or if you create objects using new. You can read more on Springs documentation: http://static.springsource.org/spring/docs/3.0.x/spring-framework-reference/html/aop.html#aop-using-aspectj
Essentially what you will do is add #Configurable annotation to the class that you want to be target of injection. You also have to enable it in Spring by having in your Spring xml. Lastly you have to decide between compile time weaving or runtime weaving. Again you can get help from spring documentation.
Loadtime weaving: http://static.springsource.org/spring/docs/3.0.x/spring-framework-reference/html/aop.html#aop-aj-ltw
If you use maven you can check this Stackoverflow question for setting up compile time AspectJ: Why doesn't AspectJ compile-time weaving of Spring's #Configurable work?
ApplicationContext.getAutowireCapableBeanFactory().autowireBean(object) will inject all dependencies into the object.

Spring AOP: Disadvantages when using it - Spring Features which use Spring AOP do not have this disadvantages?

Im working with the Spring Framework 3.0.5 and Spring Security 3.0.5 and Ive got questions to the aspect orientated programming. At the moment Im trying to figure out the disadvantages and advantages of aspect orientated programming. Of course, I know them in theory: I can avoid redundant code, I only have to make changes in the aspect, not everywhere in the code and so on. But I still got some questions:
The disadvantage I found out:
I wrote a sample application using aspects with Spring AOP. I configured an Aspect with Annotations (#Pointcut, #Before, #Aspect and so one). The methods that triggered the aspect (which were part of a Pointcut of course) were of course part of a different class and not annotated with anything.
=> I really think that one big disadvantage is that when watching those methods of the other class it was not clear that they trigger an aspect. They needed no annotations or anything else, they were just mentioned in the pointcut of the aspect. (I really hope you understand what I mean). So thats why I think that AOP makes code also less understandable!
a) Is there a solution to this problem? (Can this maybe be solved when I put the whole configuration in an XML File? I dont think so.)
b) Would this problem still exist when I would use AspectJ instead of Spring AOP?
Springs Features using Spring AOP: they dont have this disadvantage?
As Spring AOP is part of many Spring Features (just like declarative Transaction Management or (maybe) Spring Security(?)) I took a closer look at those Features. I was not able to find any disadvantage at all.
c) Lets take the declarative transaction management as an example: managing transactions is so easy with those annotations (#transactional) and I dont really find the disadvantage I mentioned above. I can see the methods that trigger specific behaviour. (all #transactional methods trigger transactional behaviour) Maybe I misunderstood something and this isnt where AOP is used? But if I did not misunderstood this, why is it here possible to see which methods trigger aspects and why isnt it possible to see in my example above? I would really like to know this!
Thank you for answering! :-)
EDIT: a) and b) are answered (use an IDE which marks those methods), c) is still missing :-)
For Point b)
If you use an Eclipse with Spring IDE and AspectJ plugin, or STS, the IDE will show you where Aspects are woven in.
For Point b)
If you use AspectJ and an IDE that supports AspectJ (Eclipse with AspectJ Plugin, or STS), then you will see markers in the souce-code where the Aspect is woven in.
An disadvantaged of Compile time AspectJ is, that you are not able to wove aspects in libraries. (without advanced techniques).
For Point c)
There is only one disadvantage of declarative Aspects like #Transactional. -- You can forget to put the annotation on the method.
But if you have for example a rule like: every public method in a Class annoteted by #Service (or if you like to build you own #TransactionalService), is transactional, then you do not need to specifiy the #Transactional annotation on each method. -- So in Summary: declarative Aspects are very good for reading (you will not overlook them), and they are good if your code is very (lets say) "individual" (instead of the term "not consistent") . But If you work in an Environment with Strong Architecural Rules (like every public method in a #Service class...), then you can Write this rules down in a Point Cut Definition, instead of using declarative Aspects.
Actually for point a) the same answer holds that Ralph gave: use Eclipse with either AspectJ plugin or if you are using Spring anyway, use STS.
That way you will see in your IDE if a certain method matches a pointcut on the left side of your editor, represented by small red arrows:
Actually, Spring AOP supports creating custom annotations.
I defined a annotation named Loggable binding with Advice.The Loggabel could be applied to any method you want.
public #interface Loggable {
}
#Aspect
public class EmployeeAnnotationAspect {
#Before("#annotation(com.ben.seal.spring.aspect.Loggable)")
public void myAdvice(){
System.out.println("Executing myAdvice!!");
}
}

Managing complexity in a dependency-injected app with a large number of beans

I'm working on an Spring application which has a large number of beans - in the hundreds - and it's getting quite cumbersome to use and document.
I'm interested in any experience you have with DI-enabled apps with a large number of beans which would aid maintainability, documentation and general usage.
Although the application is Spring-based with a couple of context files, I'm open to listening about suggestions regarding any DI container and about DI in general as well.
You can use component scan and autowiring features to dramatically decrease the amount of Spring XML configuration.
Example:
<beans>
<!-- Scans service package looking for #Service annotated beans -->
<context:component-scan base-package="my.root.package.service"/>
</beans>
Your service classes must be annotated in order to be automatically scanned:
package my.root.package.service;
#Service("fooService")
public class FooServiceImpl implements FooService{
}
You can also use the #Autowired annotation to tell Spring how to inject the bean dependencies:
package my.root.package.service;
#Service("barService")
public class BarServiceImpl implements BarService{
//Foo service injected by Spring
#Autowired
private FooService fooService;
//...
}
I found the following to be of use:
split your Spring configurations into multiple standalone configurations, and use Spring's import facility to import configuration dependencies (see here, section 3.2.2.1). That way you have a set of configurations that you can combine or disassemble as required, and they are all self-dependent (all the dependencies will be explicit and referenced)
Use an IDE that is Spring-aware, and allows you to navigate through the configurations via point-n-click on beans (references/names, to-and-from source code). Intellij works very well at this (version 7 and beyond, I think). I suspect Eclipse would do something similar.
Revise what you're injecting where. You may want to refactor multiple bean injections into one composite or 'meta' bean, or a larger component. Or you may find that components you once thought you'd need to inject have never changed, or never demanded that injectability (for testing, implementing as strategies etc.)
I used to work with a huge Spring installation, with hundreds (thousands?) of beans. Splitting the configurations up made life a lot more manageable, and simplified testing/creating standalone processes etc. But I think the Intellij Spring integration that came with Intellij made the most difference. Having a Spring-aware IDE is a major timesaver.
As #Wilson Freitas says, use autowiring. I daily work with a system that has few thousand spring managed beans using mostly autowired. But I think the notion of "retaining the overall picture" is slightly misplaced. As a system grows you can't expect to do that in the same way as you did on a smaller system. Using #Autowiring forces you to use stronger typing than xml-based spring, which again means you can use the dependency tracking features of your IDE to navigate in dependencies.
I really think it's suboptimal to think that you need to understand too much of the "full" picture when it comes to the spring configuration. You should be focusing on your code and it's dependencies. Managebility and maintainability are achieved by organizing this code well, naming things well and managing your coupling; all of the stuff that applies even if you're not using spring. Spring shouldn't change much, and with the approval of JSR-330, it may even seem like dependency injection will creep further "under the hood" of the runtime environment.
Our strategy is:
naming conventions, e.g.: fooService, fooDao, fooController;
property setters following these conventions;
autowiring by name (autowire="byName"); we had many problems with autowiring by type, especially on the controller layer

Resources