My team is moving to Spring 3.0 and there are some people who want to start moving everything into Annotations. I just get a really bad feeling in my gut (code smell?) when I see a class that has methods like this: (just an example - not all real annotations)
#Transaction
#Method("GET")
#PathElement("time")
#PathElement("date")
#Autowired
#Secure("ROLE_ADMIN")
public void manage(#Qualifier('time')int time) {
...
}
Am I just behind the times, or does this all seem like a horrible idea to anyone else? Rather then using OO concepts like inheritance and polymorphism everything is now by convention or through annotations. I just don't like it. Having to recompile all the code to change things that IMO are configuration seems wrong. But it seems to be the way everything (especially Spring) is going. Should I just "get over it" or should I push back and try to keep our code as annotation free as possible?
Actually I think that the bad feeling in your gut against has more to do with Annotations like this mixing configuration with code.
Personally I feel the same way as you do, I would prefer to leave configuration (such as transaction definitions, path elements, URLs that a controller should be mapped to, etc.) outside of the code base itself and in external Spring XML context files.
I think though that the correct approach here comes down to opinion and which method you prefer - I would predict that half the community would agree with the annotations approach and the other half would agree with the external configuration approach.
Maybe you have a problem with redundant annotations that are all over the code. With meta-annotations redundant annotations can be replaced and your annotations are at least DRY.
From the Spring Blog:
#Service
#Scope("request")
#Transactional(rollbackFor=Exception.class)
#Retention(RetentionPolicy.RUNTIME)
public #interface MyService {
}
#MyService
public class RewardsService {
…
}
Because Java evolves so slowly people are putting more features that are missing in the language into annotations. This is a good thing Java can be extended in some form and this is a bad thing as most of the annotations are some workaround and add complexity.
I was also initially skeptical about annotations, but seeing them in use, they can be a great thing. They can also be over used.
The main thing to remember about annotations is that they are static. They cannot change at runtime. Any other configuration method (xml, self-description in code, whatever) does not suffer from this. I have seen people here on SO have issues with Spring in terms of having a test environment on injecting test configurations, and having to drop down to XML to get it done.
XML isn't polymorphic, inherited or anything else either, so it is not a step backwards in that sense.
The advantage of annotations is that it can give you more static checking on your configuration and can avoid a lot of verbosity and coordination difficulties in the XML configurations (basically keeping things DRY).
Just like XML was, Annotations can be over used. The main point is to balance the needs and advantages of each. Annotations, to the degree that they give you less verbose and DRYer code, are a tool to be leveraged.
EDIT: Regarding the comment about an annotation replacing an interface or abstract class, I think that can be reasonable at the framework boundary. In a framework intended to be used by hundreds, if not thousands of projects, having an interface or base class can really crimp things (especially a base class, although if you can do it with annotations, there is no reason you couldn't do it with a regular interface.
Consider JUnit4. Before, you had to extends a base class that had a setup and tear down method. For my point, it doesn't really matter if those had been on an interface or in a base class. Now I have a completely separate project with its own inheritance hierarchy, and they all have to honor this method. First of all, they can't have their own conflicting method names (not a big deal in a testing framework, but you get my point). Second of all you have have the chain of calling super all the way down, because all methods must be coupled.
Now with JUnit4, you can have different #Before methods in different classes in the hierarchy and they can be independent of each other. There is no equally DRY way to accomplish this without annotations.
From the point of view of the developers of JUnit, it is a disaster. Much better to have a defined type that you can call setUp and teardown on. But a framework doesn't exist for the convenience of the framework developer, it exists for the convenience of the framework user.
All of this applies if your code doesn't need to care about the type (that is, in your example, nothing would every really use a Controller type anyway). Then you could even say that implementing the framework's interface is more leaky than putting on an annotation.
If, however, you are going to be writing code to read that annotation in your own project, run far away.
It's 2018 and this point is still relevant.
My biggest problem with annotations is that you don't have an idea what the annotations are doing. You're cutting some caller code off and hiding it somewhere disconnected from the callee.
Annotations were introduced to make the language more declarative and less programmatic. But if you're moving the majority of the functionality to annotations, you are effectively switching your code to a different language (and not a very good one at that). There's very little compile-time checking. This article makes the same point: https://blog.softwaremill.com/the-case-against-annotations-4b2fb170ed67
The whole heuristic of "move everything to configuration so that people don't have to learn how to code" has gotten out of control. Engineering managers aren't thinking.
Exceptions:
JUnit
JAX-RS
I personally feel that annotations have taken over too much and have blown up from their original and super useful purpose (e.g., minor things like indicating overridden method) into this crazy metaprogramming tool. I don't feel the JAva mechanism is robust enough to handle these clusters of annotations preceding each method.
For instance, I'm fighting with JUnit annotations these days because they restrict me in ways that I don't like
That being said, in my experience the XML based configuration isn't pretty either. So to quote South Park, you're choosing between a giant douche and a t*rd sandwich.
I think that the main decision you have to make is whether you are more comfortable with having a delocalization of the spring configuration (i.e., maintain two files instead of one), and whether you use tools or IDE plugins that benefit from the annotations. Another important question is whether the developers who will use or maintain your code truly understand annotations.
Like many things, there are pros and cons. In my opinion, some annotations are fine, though sometimes it feels like there is a tendency to overuse annotations when a plain old function calling approach might be superior, and taken as a whole, this can unintentionally increase cognitive load because they increase the number of ways to "do stuff."
Let me explain. For example, I'm glad you mentioned the #Transactional annotation. Most Spring developers probably are going to know about and use #Transactional. But how many of those developers know how #Transactional actually works? And would they know off the top of their head how to create and manage a transaction without using the #Transactional annotation? Using #Transactional makes it easier for me to use transactions in a majority of cases, but in particular cases when I need more fine-grained control over a transaction, it hides those details from me. So in a way it is a double edged sword.
Another example is #Profile in Spring config classes. In the general case, it makes it easier to specify which profiles you want a Spring component loaded in. However, it if you need more powerful logic than just specifying a list of profiles for which you want the component loaded, you would have to get the Environment object yourself and write a function to do this. Again, most Spring developers would probably be familiar with #Profile, but the side effect of that is they become less familiar with the details of how it works, like the Environment.acceptsProfiles(String... profiles) function, for instance.
Finally, when annotations don't work, it can be harder to understand why and you can't just put a breakpoint on the annotation. (For instance, if you forgot the #EnableTransactionManagement on your config, what would happen?) You have to find the annotation processor and debug that. With a function calling approach, you can of course just put a breakpoint in the function.
Annotations have to be used sparingly. They are good for some but not for all. At least the xml configuration approach keeps the config in one file (or multiple) instead of spread all over the place. That would introduce (as I like to call it) crappy code organization. You will never see the full picture of the configuration if it is spread across hundreds of files.
Annotations often introduce dependencies where such dependencies do not belong.
I have a class which happens by coincidence to have properties which resemble the attributes from a table in an RDBMS schema. The class was created with this mapping in mind. There is clearly a relationship between the class and the table but I am happy to keep the class free from any metadata declaring that relationship. Is it right that this class makes a reference to a table and its columns in a completely different system? I certainly don't object to external metadata that associates the two and leaves each free of an understanding of the other. What did I gain? It is not as if metadata in the source code provides type safety or mapping conformance. Any verification tool that could analyze JPA annotations could equally well analyze hibernate mapping files. Annotations did not help.
At one contract, I had created a maven module with a package of implementations of interfaces from an existing package. It is unfortunate that this new package was one of many directories within a monolithic build; I saw it as something separate from the other code. Nonetheless, the team was using classpath scanning so I had to use annotations in order to get my component wired into the system. Here I did not desire centralized configuration; I simply wanted external configuration. XML configuration was not perfect because it conflated dependency wiring with component instantiation. Given that Rod Johnson didn't believe in component based development, this was fair. Nonetheless, I felt once again that annotations did not help me.
Let's contrast this with something that doesn't bother me: TestNG and JUnit tests. I use annotations here because I write this test knowing that I am using either TestNG or JUnit. If I replace one for the other, I understand that I will have to perform a costly transition that will stray close to a rewrite of the tests.
For whatever reason, I accept that TestNG, JUnit, QUnit, unittest, and NUnit owns my test classes. Under no circumstances does either JPA or Hibernate own those domain classes which happen to get mapped to tables. Under no circumstances does Spring own my services. I control my logical and physical packaging in order to isolate units which depend upon either. I want to ensure that a move away from one doesn't leave me crippled because of all the dependencies it left behind. Saying goodbye is always easier than leaving. At some point, leaving is necessary.
Check these answers to similar questions
What are the Pros/Cons of Annotations (non-compiler) compared to xml config files
Xml configuration versus Annotation based configuration
Basically it boils down to: Use both. Both of them have there usecases. Don't use annotations for things which should remain configurable without recompiling everything (especially things which maybe your user should be able to configure without needing you to recompile all)
I think it depends to some extent on when you started programming. Personally, I think they are horrid. Primarily because they have some quasi-'meaning' which you will not understand unless you happen to be aware of the annotation in question. As such they form a new programming language all by themselves and move you further away from POJOs. Compared to (say) plain old OO code. Second reason - they can prevent the compiler doing your work for you. If I have a large code base and want to refactor something or rename something I'd ideally like the compiler to throw up everything that needs to be changed, or as much as possible. An annotation should just be that. An annotation. Not central to the behaviour of your code. They were designed originally to be optionally omitted upon compilation which tells you all you need to know.
And yes, I am aware that XML config suffers in the same way. That doesn't make it worse, just equally bad. At least I can pretend to ignore that though - it doesn't stare me in the face in every single method or parameter declaration.
Given the choice I'd actually prefer the horrible old J2EE remote/home interfaces etc (so criticised by the Spring folks originally) as at least that gives me an idea of whats happening without having to research #CoolAidFrameworkThingy and its foibles.
One of the problems with the framework folks is that they need to tie you to their framework in order to make the whole enterprise financially viable. This is at odds with designing a framework well (i.e. for it to be as independant and removeable from your code as possible).
Unfortunately, though, annotations are trendy. So you will have a hard time preventing your team using them unless you are into code reviews/standards and the like (also, out of fashion!)
I read that Stroustup left annotations out of C++ as he feared they would be mis-used. Sometimes things go in the wrong direction for decades, but you can hope things will come full circle in time..
I think annotations are good if they are used with measure. Annotations like #WebService do a lot of work at deployment and run time, but they don't interfere in the class. #Cachexxx or #Transactional clearly interfere by creating proxies and a lot of artifacts, but I think they are under control.
Thing begin to mess when using Hibernate or JPA with annotations and CDI. Annotations grow a lot.
IMO #Service and #Repository are interferences of Spring in your application code. They make your application Spring dependant and only for Spring use.
The case of Spring Data Graph is another story. #NodeEntity, for instance, add methods to the class at build time to save the domain object. Unless you have Eclipse and Spring plugin you will errors because those methods don't exist in source code.
Configuration near the object has its benefits, but also a single configuration point. Annotations are good with measure, but they aren't good for everything, and definitively bad when there are as much annotation lines as source code lines.
I think the path Spring is going is wrong; mainly because in some cases there is no other way to do such funny things. It's is as if Spring wants to do xtreme coding, and at the same time they lock developers into Spring framework. Probably Java language needs another way to do some things.
Annotations are plain bad in my experience:
Inability to enforce type safety in annotations
Serialization issues
Cross compiling (to for instance javascript) can be an issue.
Libraries/frameworks requiring annotations exclude non-annotated classes from external libraries.
not overridable or interchangeable
your projects eventually becomes strongly dependant on the system that requires the annotations
If Java would have something like "method literals" you could annotate a class in a corresponding annotation class.
Something like as following:
Take for instance javax.persistence, and the following annotated class:
#Entity
class Person
{
#Column
private String firstname;
public String getFirstname() { return firstname; }
public void setFirstname(String value) { firstname = value; }
#Column
private String surname;
public String getSurname() { return surname; }
public void setSurname(String value) { surname = value; }
}
Instead of the annotations, I'd suggest a mapping class like:
class PersonEntity extends Entity<Person> {
#Override
public Class<Person> getEntityClass() { return Person.class;}
#Override
public Collection<PersistentProperty> getPersistentProperties() {
LinkedList<PersistentProperty> result = new LinkedList<>();
result.add(new PersistentProperty<Person>(Person#getFirstname, Person#setFirstname);
result.add(new PersistentProperty<Person>(Person#getSurname, Person#setSurname);
return result;
}
}
The fictional "#" sign in this pseudo java code represents a method literal, which, when invoked on an instance of the given class, invokes the corresponding delegate (signed with "::" since java 8) of that instance.
The "PersistentProperty" class should be able to enforce the method literals to be referring to the given generic argument, in this case the class Person.
This way, you have more benefits than annotations can deliver (like subclassing your 'annotate'-class) and you have none of the aforementioned cons.
You can have more domain-specific approaches too.
The only pre annotations have over this, is that with annotations you can quickly see whether you have forgotten to include a property/method. But this too can be handled more concise and more correct with better metadata support in Java (think for instance of something like required/optional like in Protocolbuffers)
Related
I'm fairly new to Java and joining a project that leverages the DDD pattern (supposedly). I come from a strong python background and am fairly anal about unit test driven design. That said, one of the challenges of moving to Java is the testability of Service layers.
Our REST-like project stack is laid out as follows:
ServiceHandlers which handles request/response, etc and calls specific implementations of IService (eg. DocumentService)
DocumentService - handles auditing, permission checking, etc with methods such as makeOwner(session, user, doc)
Currently, something like DocumentService has repository dependencies injected via guice. In a public method like DocumentService.makeOwner, we want to ensure the session user is an admin as well as check if the target user is already an owner (leveraging the injected repositories). This results in some dupe code - one for both users involved to resolve the user and ensure membership, permissions, etc etc. To eliminate this redundant code, I want make a sort of super simpleisOwner(user, doc) call that I can concisely mock out for various test scenarios (such as throwing the exception when the user can't be resolved, etc). Here is where my googling fails me.
If I put this in the same class as DocumentService, I can't mock it while testing makeOwner in the same class (due to Mockito limitations) even though it somewhat feels like it should go here (option1).
If I put it in a lower class like DocumentHelpers, it feels slightly funny but I can easily mock it out. Also, DocumentHelpers needs the injected repository as well, which is fine with guice. (option 2)
I should add that there are numerous spots of this nature in our infant code base that are untestable currently because methods are non-statically calling helper-like methods in the same *Service class not used by the upper ServiceHandler class. However, at this stage, I can't tell if this is poor design or just fine.
So I ask more experienced Java developers:
Does introducing "Service Helpers" seem like a valid solution?
Is this counter to DDD principals?
If not, is there are more DDD-friendly naming convention for this aside from "Helpers"?
3 bits to add:
My googling has mostly come up with debates over "helpers" as static utility methods for stateless operations like date formatting, which doesn't fit my issue.
I don't want to use PowerMock since it breaks code coverage and is pretty ugly to use.
In python I'd probably call the "Service Helper" layer described above as internal_api, but that seems to have a different meaning in Java, especially since I need the classes to be public to unit test them.
Any guidance is appreciated.
That the user who initiates the action must be an admin looks like an application-level access control concern. DDD doesn't have much of an opinion about how you should do that. For testability and separation of concerns purposes, it might be a better idea to have some kind of separate non-static class than a method in the same service or a static helper though.
Checking that the future owner is already an owner (if I understand correctly) might be a different animal. It could be an invariant in your domain. If so, the preferred way is to rely on an Aggregate to enforce that rule. However, it's not clear from your description whether Document is an aggregate and if it or another aggregate contains the data needed to tell if a user is owner.
Alternatively, you could verify the rule at the Application layer level but it means that your domain model could go inconsistent if the state change is triggered by something else than that Application layer.
As I learn more about DDD, my question doesn't seem to be all that DDD related and more just about general hierarchy of the code structure and interactions of the layers. We ended up going with a separate DocumentServiceHelpers class that could be mocked out. This contains methods like isOwner that we can mock to return true or false as needed to test our DocumentService handling more easily. Thanks to everyone for playing along.
I am trying to have an annotation #FeatureDependent be used on methods to signal that the method requires certain things to be enabled in order for it to work. And I was wondering if it was possible to have a method called everytime a method with #FeatureDependent was called which would check if the criteria were met for the method to be called.
It sounds like you are describing Aspect Oriented Programming (AOI). This technique allows you to address "cross-cutting" concerns, tasks like logging, security, and transaction management which tend to affect many methods in the same manner. Your use case sounds like it would be a good fit for AOP.
There are two common approaches to AOP. The first mechanism is to create objects in a container (e.g. a Spring container). The container can then scan the class, detect any advice that needs to be applied, and apply the advice via dynamic proxies (Googling Spring and AOP is a good place to start with this). The downside is that your components will need to be constructed by a container so it makes sense for larger components.
The second approach is an extra compilation step (sometimes done at compilation, sometimes done as a separate compilation phase, and sometimes done by a weaving class loader) to wire in the additional methods. This is typically called "weaving" and AspectJ is a common library to look into for this.
Both approaches will allow you to apply "advice" (code run before and after a method invocation) based on annotations on an object. Explaining either in more detail would be beyond the scope of a SO answer but I hope it can get you started.
I should warn that AOP has gotten a bit of a reputation for complicating the flow of an application and tending to be difficult to understand and debug. As a result it has declined in popularity lately.
Another approach is to use something like Servlet Filters, basically a single choke point for all requests and workflows where you can apply various logging & security mechanisms. Such an approach tends to be a little easier to understand and involve a bit less "black magic".
I would like to build my own custom DI framework based on Java annotations and I need a little direction to get started. I know it would be much easier to use one of the many wonderful frameworks out there such as guice or spring, but for the sake of my own curiosity, i'd like to build my own.
I'm not very familiar with annotations, so i'm having a bit of trouble finding resources and would really appreciate someone just sort of spelling out a few of the steps i'll need to take to get started.
As fore mentioned, id like to take a factory approach and somehow label my getters with an #Resource or #Injectable type annotation, and then in my business classes be able to set my variable dependencies with an #Inject annotation and have the resource automatically available.
Does anyone have any sort of resource they can pass along to help me understand the process of tagging methods based on annotations and then retrieving values from a separate class based on an annotation. A little direction is all I need, something to get me started. And of course i'll be happy to post a little code sample here once I get going, for the sake of others future reading of course.
EDIT
The resources I am using to put this together:
Java Reflection: Annotations
How to find annotations in a given package: Stack Overflow ?
Scanning Annotations at Runtime
I have not actually finished writing this yet, but the basic task list is going to be as follows (for anyone who might be interested in doing something similar in the future)
At class runtime scan for all #Inject fields and get object type.
Scan all classes (or just a specific package of classes (I haven't
decided yet)) for annotated methods #InjectableResource.
Loop all annotated methods and find the method that returns the
object type I am looking for.
Run the method and get the dependency.
It will also be helpful to note that when scanning all the classes I will be using a library called Javassist. Basically what this does is allows me to read the bytecode information of each class without actually loading the class. So I can read the annotation strings without creating serious memory problems.
Interesting that you want to build your own. I love Google Guice - it makes code so elegant and simple.
I've used this guide before which I found pretty useful for learning about annotations and how you can pull them out of classes and methods.
You will have to define your own Annotations which is done using #interface. Then you will have to define some kind of class for doing bindings e.g. where you see an interface bind in this concrete class. Finally, you will need some logic to pull it altogether e.g. go through each class, find each annotation, and then find a suitable binding.
Give consideration to things like lazy instantiation through Reflections and singletons. Guice, for example, allows you to use a singleton so your only using one instance of the concrete class, or you can bind a new version each time.
Good luck!
Have a look at the following methods:
java/lang/Class.html#getAnnotation(java.lang.Class)
java/lang/Class.html#getAnnotations()
java/lang/Class.html#getDeclaredAnnotations()
Methods of the same name also exist for the java/lang/reflect/Method, java/lang/reflect/Field and java/lang/reflect/Constructor classes.
So in order to use these sorts of methods, you need to know a bit about Java reflection.
As you many know when you proxy an object, like when you create a bean with transactional attributes for Spring/EJB or even when you create a partial mock with some frameworks, the proxies object doesn't know that, and internal calls are not redirected, and then not intercepted either...
That's why if you do something like that in Spring:
#Transactionnal
public void doSomething() {
doSomethingInNewTransaction();
doSomethingInNewTransaction();
doSomethingInNewTransaction();
}
#Transactional(propagation = Propagation.REQUIRES_NEW)
public void doSomethingInNewTransaction() {
...
}
When you call doSomething, you expect to have 3 new transactions in addition to the main one, but actually, due to this problem you only get one...
So i wonder how do you do to handle these kind of problems...
I'm actually in a situation where i must handle a complex transactional system, and i don't see any better way than splitting my service into many small services, so that I'm sure to pass through all the proxies...
That bothers me a lot because all the code belongs to the same functional domain and should not be split...
I've found this related question with interesting answers:
Spring - #Transactional - What happens in background?
Rob H says that we can inject the spring proxy inside the service, and call proxy.doSomethingInNewTransaction(); instead.
It's quite easy to do and it works, but i don't really like it...
Yunfeng Hou says this:
So I write my own version of CglibSubclassingInstantiationStrategy and
proxy creator so that it will use CGLIB to generate a real subclass
which delegates call to its super rather than another instance, which
Spring is doing now. So I can freely annotate on any methods(as long
as it is not private), and from wherever I call these methods, they
will be taken care of. Well, I still have price to pay: 1. I must list
all annotations that I want to enable the new CGLIB sub class
creation. 2. I can not annotate on a final method since I am now
generating subclass, so a final method can not be intercepted.
What does he mean by "which spring is doing now"? Does this mean internal transactional calls are now intercepted?
What do you think is better?
Do you split your classes when you need some transactional granularity?
Or do you use some workaround like above? (please share it)
I'll talk about Spring and #Transactional but the advise applies for many other frameworks also.
This is an inherent problem with proxy based aspects. It is discussed in the spring documentation here:
http://static.springsource.org/spring/docs/3.0.x/spring-framework-reference/html/aop.html#aop-understanding-aop-proxies
There are a number of possible solutions.
Refactor your classes to avoid the self-invocation calls that bypass the proxy.
The Spring documentation describes this as "The best approach (the term best is used loosely here)".
Advantages of this approach are its simplicity and that there are no ties to any framework. However, it may not be appropriate for a very transactional heavy code base as you'd end up with many trivially small classes.
Internally in the class get a reference to the proxy.
This can be done by injecting the proxy or with hard coded " AopContext.currentProxy()" call (see Spring docs above.).
This method allows you to avoid splitting the classes but in many ways negates the advantages of using the transactional annotation. My personal opinion is that this is one of those things that is a little ugly but the ugliness is self contained and might be the pragmatic approach if lots of transactions are used.
Switch to using AspectJ
As AspectJ does not use proxies then self-invocation is not a problem
This is a very clean method though - it is at the expense of introducing another framework. I've worked on a large project where AspectJ was introduced for this very reason.
Don't use #Transactional at all
Refactor your code to use manual transaction demarcation - possibly using the decorator pattern.
An option - but one that requires moderate refactoring, introducing additional framework ties and increased complexity - so probably not a preferred option
My Advice
Usually splitting up the code is the best answer and can also be good thing for seperation of concerns also. However, if I had a framework/application that heavily relied on nested transactions I would consider using AspectJ to allow self-invocation.
As always when modelling and designing complex use cases - focus on understandable and maintainable design and code. If you prefer a certain pattern or design but it clashes with the underlying framework, consider if it's worth a complex workaround to shoehorn your design into the framework, or if you should compromise and conform your design to the framework where necessary. Don't fight the framework unless you absolutely have to.
My advice - if you can accomplish your goal with such an easy compromise as to split out into a few extra service classes - do it. It sounds a lot cheaper in terms of time, testing and agony than the alternative. And it sure sounds a lot easier to maintain and less of a headache for the next guy to take over.
I usually make it simple, so I split the code into two objects.
The alternative is to demarcate the new transaction yourself, if you need to keep everything in the same file, using a TransactionTemplate. A few more lines of code, but not more than defining a new bean. And it sometimes makes the point more obvious.
I'm thinking of annotating my Domain Objects. This will ease manipulation of the DOs. Keeping the domain code free from other external stuff is also important for me.
Any comment on the "corruption" of the domain code by adding annotations?
Are you for/against adding annotation to Domain Objects, and why?
I think annotating if it makes the code simpler is a good idea, but, you should look at what is already out there, and at least use what may be a standard for your annotation names. For example you can look at JDBC 4.0, in Java (http://onjava.com/pub/a/onjava/2006/08/02/jjdbc-4-enhancements-in-java-se-6.html?page=2), or Spring as examples.
This will do two things. One is that if you decide to move to using these annotations at some point, and get rid of your own, then your code won't change. Two, it shortens the learning curve for others.
Your annotations may not be going to the database, but there are numerous annotation models out there, just be certain if you create your own new names that you are doing something sufficiently unique otherwise it justs gets confusing for those that need to read your code.
Annotations (like most anything) have tradeoffs. The big one is that they are static. If you want to change during runtime a property represented in an annotation, you are out of luck.
They can get a little difficult to work with when you get into involved scenarios (especially when you deal with annotated annotations).
And if you have a lot of them, they can tend to make the code unreadable.
However, in moderation, kept simple and done right, they can really make code and configuration a lot simpler and cleaner.
Have a look at Terracotta - very possibly you don't have to write your own annotations. We were presented with similar dilemma (our DOs weren't intended for relation db) and Terracotta turned out to be a real life savior
We use annotations for specific things - i.e. special handling of Strings and the like - and it works like a charm. Annotations are a great way of handling "metadata" kind of information - data about the data object. I would recommend looking at the current J2EE annotations (I think it is version 5.0?) as that is used by most ORM systems (i.e. Hibernate and the like).
I prefer my annotations to be descriptive rather than functional. For example, the JCIP concurrency annotations describe information about a class, but don't provide functionality in an of themselves. Annotations that cause functionality tend to be PFM (pure effing magic) and make code more difficult to understand.
That's not a hard rule, but it's a pain when annotations do some of the functional configuration and configuration files (like XML) handle other configuration. It leads to code that requires you to look all over the place and understand multiple configuration schemes for how things are supposed to work.