Aspect-Oriented Programming (AOP) complements Object-Oriented Programming (OOP) by providing another way of thinking about program structure. The key unit of modularity in OOP is the class, whereas in AOP the unit of modularity is the aspect. Aspects enable the modularization of concerns such as transaction management that cut across multiple types and objects. (Such concerns are often termed crosscutting concerns in AOP literature.)
One of the key components of Spring is the AOP framework. While the Spring IoC container does not depend on AOP, meaning you do not need to use AOP if you don’t want to, AOP complements Spring IoC to provide a very capable middleware solution.
AOP is used in the Spring Framework to…
Note | |
---|---|
If you are interested only in generic declarative services or other pre-packaged declarative middleware services such as pooling, you do not need to work directly with Spring AOP, and can skip most of this chapter. |
Let us begin by defining some central AOP concepts and terminology. These terms are not Spring-specific… unfortunately, AOP terminology is not particularly intuitive; however, it would be even more confusing if Spring used its own terminology.
@Aspect
annotation (the @AspectJ
style).
IsModified
interface, to simplify caching. (An introduction is known as an
inter-type declaration in the AspectJ community.)
Types of advice:
Around advice is the most general kind of advice. Since Spring AOP, like AspectJ,
provides a full range of advice types, we recommend that you use the least powerful
advice type that can implement the required behavior. For example, if you need only to
update a cache with the return value of a method, you are better off implementing an
after returning advice than an around advice, although an around advice can accomplish
the same thing. Using the most specific advice type provides a simpler programming model
with less potential for errors. For example, you do not need to invoke the proceed()
method on the JoinPoint
used for around advice, and hence cannot fail to invoke it.
In Spring 2.0, all advice parameters are statically typed, so that you work with advice
parameters of the appropriate type (the type of the return value from a method execution
for example) rather than Object
arrays.
The concept of join points, matched by pointcuts, is the key to AOP which distinguishes it from older technologies offering only interception. Pointcuts enable advice to be targeted independently of the Object-Oriented hierarchy. For example, an around advice providing declarative transaction management can be applied to a set of methods spanning multiple objects (such as all business operations in the service layer).
Spring AOP is implemented in pure Java. There is no need for a special compilation process. Spring AOP does not need to control the class loader hierarchy, and is thus suitable for use in a Servlet container or application server.
Spring AOP currently supports only method execution join points (advising the execution of methods on Spring beans). Field interception is not implemented, although support for field interception could be added without breaking the core Spring AOP APIs. If you need to advise field access and update join points, consider a language such as AspectJ.
Spring AOP’s approach to AOP differs from that of most other AOP frameworks. The aim is not to provide the most complete AOP implementation (although Spring AOP is quite capable); it is rather to provide a close integration between AOP implementation and Spring IoC to help solve common problems in enterprise applications.
Thus, for example, the Spring Framework’s AOP functionality is normally used in conjunction with the Spring IoC container. Aspects are configured using normal bean definition syntax (although this allows powerful "autoproxying" capabilities): this is a crucial difference from other AOP implementations. There are some things you cannot do easily or efficiently with Spring AOP, such as advise very fine-grained objects (such as domain objects typically): AspectJ is the best choice in such cases. However, our experience is that Spring AOP provides an excellent solution to most problems in enterprise Java applications that are amenable to AOP.
Spring AOP will never strive to compete with AspectJ to provide a comprehensive AOP solution. We believe that both proxy-based frameworks like Spring AOP and full-blown frameworks such as AspectJ are valuable, and that they are complementary, rather than in competition. Spring seamlessly integrates Spring AOP and IoC with AspectJ, to enable all uses of AOP to be catered for within a consistent Spring-based application architecture. This integration does not affect the Spring AOP API or the AOP Alliance API: Spring AOP remains backward-compatible. See the following chapter for a discussion of the Spring AOP APIs.
Note | |
---|---|
One of the central tenets of the Spring Framework is that of non-invasiveness; this is the idea that you should not be forced to introduce framework-specific classes and interfaces into your business/domain model. However, in some places the Spring Framework does give you the option to introduce Spring Framework-specific dependencies into your codebase: the rationale in giving you such options is because in certain scenarios it might be just plain easier to read or code some specific piece of functionality in such a way. The Spring Framework (almost) always offers you the choice though: you have the freedom to make an informed decision as to which option best suits your particular use case or scenario. One such choice that is relevant to this chapter is that of which AOP framework (and which AOP style) to choose. You have the choice of AspectJ and/or Spring AOP, and you also have the choice of either the @AspectJ annotation-style approach or the Spring XML configuration-style approach. The fact that this chapter chooses to introduce the @AspectJ-style approach first should not be taken as an indication that the Spring team favors the @AspectJ annotation-style approach over the Spring XML configuration-style. See Section 9.4, “Choosing which AOP declaration style to use” for a more complete discussion of the whys and wherefores of each style. |
Spring AOP defaults to using standard JDK dynamic proxies for AOP proxies. This enables any interface (or set of interfaces) to be proxied.
Spring AOP can also use CGLIB proxies. This is necessary to proxy classes, rather than interfaces. CGLIB is used by default if a business object does not implement an interface. As it is good practice to program to interfaces rather than classes, business classes normally will implement one or more business interfaces. It is possible to force the use of CGLIB, in those (hopefully rare) cases where you need to advise a method that is not declared on an interface, or where you need to pass a proxied object to a method as a concrete type.
It is important to grasp the fact that Spring AOP is proxy-based. See Section 9.6.1, “Understanding AOP proxies” for a thorough examination of exactly what this implementation detail actually means.
@AspectJ refers to a style of declaring aspects as regular Java classes annotated with annotations. The @AspectJ style was introduced by the AspectJ project as part of the AspectJ 5 release. Spring interprets the same annotations as AspectJ 5, using a library supplied by AspectJ for pointcut parsing and matching. The AOP runtime is still pure Spring AOP though, and there is no dependency on the AspectJ compiler or weaver.
Note | |
---|---|
Using the AspectJ compiler and weaver enables use of the full AspectJ language, and is discussed in Section 9.8, “Using AspectJ with Spring applications”. |
To use @AspectJ aspects in a Spring configuration you need to enable Spring support for configuring Spring AOP based on @AspectJ aspects, and autoproxying beans based on whether or not they are advised by those aspects. By autoproxying we mean that if Spring determines that a bean is advised by one or more aspects, it will automatically generate a proxy for that bean to intercept method invocations and ensure that advice is executed as needed.
The @AspectJ support can be enabled with XML or Java style configuration. In either
case you will also need to ensure that AspectJ’s aspectjweaver.jar
library is on the
classpath of your application (version 1.6.8 or later). This library is available in the
'lib'
directory of an AspectJ distribution or via the Maven Central repository.
To enable @AspectJ support with Java @Configuration
add the @EnableAspectJAutoProxy
annotation:
@Configuration @EnableAspectJAutoProxy public class AppConfig { }
To enable @AspectJ support with XML based configuration use the aop:aspectj-autoproxy
element:
<aop:aspectj-autoproxy/>
This assumes that you are using schema support as described in Chapter 34, XML Schema-based configuration. See Section 34.2.7, “the aop schema” for how to import the tags in the aop namespace.
With the @AspectJ support enabled, any bean defined in your application context with a
class that is an @AspectJ aspect (has the @Aspect
annotation) will be automatically
detected by Spring and used to configure Spring AOP. The following example shows the
minimal definition required for a not-very-useful aspect:
A regular bean definition in the application context, pointing to a bean class that has
the @Aspect
annotation:
<bean id="myAspect" class="org.xyz.NotVeryUsefulAspect"> <!-- configure properties of aspect here as normal --> </bean>
And the NotVeryUsefulAspect
class definition, annotated with
org.aspectj.lang.annotation.Aspect
annotation;
package org.xyz; import org.aspectj.lang.annotation.Aspect; @Aspect public class NotVeryUsefulAspect { }
Aspects (classes annotated with @Aspect
) may have methods and fields just like any
other class. They may also contain pointcut, advice, and introduction (inter-type)
declarations.
Autodetecting aspects through component scanning | |
---|---|
You may register aspect classes as regular beans in your Spring XML configuration, or autodetect them through classpath scanning - just like any other Spring-managed bean. However, note that the @Aspect annotation is not sufficient for autodetection in the classpath: For that purpose, you need to add a separate @Component annotation (or alternatively a custom stereotype annotation that qualifies, as per the rules of Spring’s component scanner). |
Advising aspects with other aspects? | |
---|---|
In Spring AOP, it is not possible to have aspects themselves be the target of advice from other aspects. The @Aspect annotation on a class marks it as an aspect, and hence excludes it from auto-proxying. |
Recall that pointcuts determine join points of interest, and thus enable us to control
when advice executes. Spring AOP only supports method execution join points for Spring
beans, so you can think of a pointcut as matching the execution of methods on Spring
beans. A pointcut declaration has two parts: a signature comprising a name and any
parameters, and a pointcut expression that determines exactly which method
executions we are interested in. In the @AspectJ annotation-style of AOP, a pointcut
signature is provided by a regular method definition, and the pointcut expression is
indicated using the @Pointcut
annotation (the method serving as the pointcut signature
must have a void
return type).
An example will help make this distinction between a pointcut signature and a pointcut
expression clear. The following example defines a pointcut named 'anyOldTransfer'
that
will match the execution of any method named 'transfer'
:
@Pointcut("execution(* transfer(..))")// the pointcut expression private void anyOldTransfer() {}// the pointcut signature
The pointcut expression that forms the value of the @Pointcut
annotation is a regular
AspectJ 5 pointcut expression. For a full discussion of AspectJ’s pointcut language, see
the AspectJ
Programming Guide (and for extensions, the
AspectJ 5
Developers Notebook) or one of the books on AspectJ such as "Eclipse AspectJ" by Colyer
et. al. or "AspectJ in Action" by Ramnivas Laddad.
Spring AOP supports the following AspectJ pointcut designators (PCD) for use in pointcut expressions:
Because Spring AOP limits matching to only method execution join points, the discussion
of the pointcut designators above gives a narrower definition than you will find in the
AspectJ programming guide. In addition, AspectJ itself has type-based semantics and at
an execution join point both this
and target
refer to the same object - the
object executing the method. Spring AOP is a proxy-based system and differentiates
between the proxy object itself (bound to this
) and the target object behind the
proxy (bound to target
).
Note | |
---|---|
Due to the proxy-based nature of Spring’s AOP framework, protected methods are by definition not intercepted, neither for JDK proxies (where this isn’t applicable) nor for CGLIB proxies (where this is technically possible but not recommendable for AOP purposes). As a consequence, any given pointcut will be matched against public methods only! If your interception needs include protected/private methods or even constructors, consider the use of Spring-driven native AspectJ weaving instead of Spring’s proxy-based AOP framework. This constitutes a different mode of AOP usage with different characteristics, so be sure to make yourself familiar with weaving first before making a decision. |
Spring AOP also supports an additional PCD named bean
. This PCD allows you to limit
the matching of join points to a particular named Spring bean, or to a set of named
Spring beans (when using wildcards). The bean
PCD has the following form:
bean(idOrNameOfBean)
The idOrNameOfBean
token can be the name of any Spring bean: limited wildcard
support using the *
character is provided, so if you establish some naming
conventions for your Spring beans you can quite easily write a bean
PCD expression
to pick them out. As is the case with other pointcut designators, the bean
PCD can
be &&'ed, ||'ed, and ! (negated) too.
Note | |
---|---|
Please note that the The |
Pointcut expressions can be combined using &&, || and !. It is also possible to
refer to pointcut expressions by name. The following example shows three pointcut
expressions: anyPublicOperation
(which matches if a method execution join point
represents the execution of any public method); inTrading
(which matches if a method
execution is in the trading module), and tradingOperation
(which matches if a method
execution represents any public method in the trading module).
@Pointcut("execution(public * *(..))") private void anyPublicOperation() {} @Pointcut("within(com.xyz.someapp.trading..*)") private void inTrading() {} @Pointcut("anyPublicOperation() && inTrading()") private void tradingOperation() {}
It is a best practice to build more complex pointcut expressions out of smaller named components as shown above. When referring to pointcuts by name, normal Java visibility rules apply (you can see private pointcuts in the same type, protected pointcuts in the hierarchy, public pointcuts anywhere and so on). Visibility does not affect pointcut matching.
When working with enterprise applications, you often want to refer to modules of the application and particular sets of operations from within several aspects. We recommend defining a "SystemArchitecture" aspect that captures common pointcut expressions for this purpose. A typical such aspect would look as follows:
package com.xyz.someapp; import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.Pointcut; @Aspect public class SystemArchitecture { /** * A join point is in the web layer if the method is defined * in a type in the com.xyz.someapp.web package or any sub-package * under that. */ @Pointcut("within(com.xyz.someapp.web..*)") public void inWebLayer() {} /** * A join point is in the service layer if the method is defined * in a type in the com.xyz.someapp.service package or any sub-package * under that. */ @Pointcut("within(com.xyz.someapp.service..*)") public void inServiceLayer() {} /** * A join point is in the data access layer if the method is defined * in a type in the com.xyz.someapp.dao package or any sub-package * under that. */ @Pointcut("within(com.xyz.someapp.dao..*)") public void inDataAccessLayer() {} /** * A business service is the execution of any method defined on a service * interface. This definition assumes that interfaces are placed in the * "service" package, and that implementation types are in sub-packages. * * If you group service interfaces by functional area (for example, * in packages com.xyz.someapp.abc.service and com.xyz.someapp.def.service) then * the pointcut expression "execution(* com.xyz.someapp..service.*.*(..))" * could be used instead. * * Alternatively, you can write the expression using the 'bean' * PCD, like so "bean(*Service)". (This assumes that you have * named your Spring service beans in a consistent fashion.) */ @Pointcut("execution(* com.xyz.someapp..service.*.*(..))") public void businessService() {} /** * A data access operation is the execution of any method defined on a * dao interface. This definition assumes that interfaces are placed in the * "dao" package, and that implementation types are in sub-packages. */ @Pointcut("execution(* com.xyz.someapp.dao.*.*(..))") public void dataAccessOperation() {} }
The pointcuts defined in such an aspect can be referred to anywhere that you need a pointcut expression. For example, to make the service layer transactional, you could write:
<aop:config> <aop:advisor pointcut="com.xyz.someapp.SystemArchitecture.businessService()" advice-ref="tx-advice"/> </aop:config> <tx:advice id="tx-advice"> <tx:attributes> <tx:method name="*" propagation="REQUIRED"/> </tx:attributes> </tx:advice>
The <aop:config>
and <aop:advisor>
elements are discussed in Section 9.3, “Schema-based AOP support”. The
transaction elements are discussed in Chapter 12, Transaction Management.
Spring AOP users are likely to use the execution
pointcut designator the most often.
The format of an execution expression is:
execution(modifiers-pattern? ret-type-pattern declaring-type-pattern? name-pattern(param-pattern)
throws-pattern?)
All parts except the returning type pattern (ret-type-pattern in the snippet above),
name pattern, and parameters pattern are optional. The returning type pattern determines
what the return type of the method must be in order for a join point to be matched. Most
frequently you will use *
as the returning type pattern, which matches any return
type. A fully-qualified type name will match only when the method returns the given
type. The name pattern matches the method name. You can use the *
wildcard as all or
part of a name pattern. The parameters pattern is slightly more complex: ()
matches a
method that takes no parameters, whereas (..)
matches any number of parameters (zero
or more). The pattern (*)
matches a method taking one parameter of any type,
(*,String)
matches a method taking two parameters, the first can be of any type, the
second must be a String. Consult the
Language
Semantics section of the AspectJ Programming Guide for more information.
Some examples of common pointcut expressions are given below.
execution(public * *(..))
execution(* set*(..))
AccountService
interface:
execution(* com.xyz.service.AccountService.*(..))
execution(* com.xyz.service.*.*(..))
execution(* com.xyz.service..*.*(..))
within(com.xyz.service.*)
within(com.xyz.service..*)
AccountService
interface:
this(com.xyz.service.AccountService)
Note | |
---|---|
this is more commonly used in a binding form :- see the following section on advice for how to make the proxy object available in the advice body. |
AccountService
interface:
target(com.xyz.service.AccountService)
Note | |
---|---|
target is more commonly used in a binding form :- see the following section on advice for how to make the target object available in the advice body. |
Serializable
:
args(java.io.Serializable)
Note | |
---|---|
args is more commonly used in a binding form :- see the following section on advice for how to make the method arguments available in the advice body. |
Note that the pointcut given in this example is different to execution(*
*(java.io.Serializable))
: the args version matches if the argument passed at runtime is
Serializable, the execution version matches if the method signature declares a single
parameter of type Serializable
.
@Transactional
annotation:
@target(org.springframework.transaction.annotation.Transactional)
Note | |
---|---|
@target can also be used in a binding form :- see the following section on advice for how to make the annotation object available in the advice body. |
@Transactional
annotation:
@within(org.springframework.transaction.annotation.Transactional)
Note | |
---|---|
@within can also be used in a binding form :- see the following section on advice for how to make the annotation object available in the advice body. |
@Transactional
annotation:
@annotation(org.springframework.transaction.annotation.Transactional)
Note | |
---|---|
@annotation can also be used in a binding form :- see the following section on advice for how to make the annotation object available in the advice body. |
@Classified
annotation:
@args(com.xyz.security.Classified)
Note | |
---|---|
@args can also be used in a binding form :- see the following section on advice for how to make the annotation object(s) available in the advice body. |
tradeService
:
bean(tradeService)
*Service
:
bean(*Service)
During compilation, AspectJ processes pointcuts in order to try and optimize matching performance. Examining code and determining if each join point matches (statically or dynamically) a given pointcut is a costly process. (A dynamic match means the match cannot be fully determined from static analysis and a test will be placed in the code to determine if there is an actual match when the code is running). On first encountering a pointcut declaration, AspectJ will rewrite it into an optimal form for the matching process. What does this mean? Basically pointcuts are rewritten in DNF (Disjunctive Normal Form) and the components of the pointcut are sorted such that those components that are cheaper to evaluate are checked first. This means you do not have to worry about understanding the performance of various pointcut designators and may supply them in any order in a pointcut declaration.
However, AspectJ can only work with what it is told, and for optimal performance of matching you should think about what they are trying to achieve and narrow the search space for matches as much as possible in the definition. The existing designators naturally fall into one of three groups: kinded, scoping and context:
A well written pointcut should try and include at least the first two types (kinded and scoping), whilst the contextual designators may be included if wishing to match based on join point context, or bind that context for use in the advice. Supplying either just a kinded designator or just a contextual designator will work but could affect weaving performance (time and memory used) due to all the extra processing and analysis. Scoping designators are very fast to match and their usage means AspectJ can very quickly dismiss groups of join points that should not be further processed - that is why a good pointcut should always include one if possible.
Advice is associated with a pointcut expression, and runs before, after, or around method executions matched by the pointcut. The pointcut expression may be either a simple reference to a named pointcut, or a pointcut expression declared in place.
Before advice is declared in an aspect using the @Before
annotation:
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.Before; @Aspect public class BeforeExample { @Before("com.xyz.myapp.SystemArchitecture.dataAccessOperation()") public void doAccessCheck() { // ... } }
If using an in-place pointcut expression we could rewrite the above example as:
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.Before; @Aspect public class BeforeExample { @Before("execution(* com.xyz.myapp.dao.*.*(..))") public void doAccessCheck() { // ... } }
After returning advice runs when a matched method execution returns normally. It is
declared using the @AfterReturning
annotation:
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.AfterReturning; @Aspect public class AfterReturningExample { @AfterReturning("com.xyz.myapp.SystemArchitecture.dataAccessOperation()") public void doAccessCheck() { // ... } }
Note | |
---|---|
Note: it is of course possible to have multiple advice declarations, and other members as well, all inside the same aspect. We’re just showing a single advice declaration in these examples to focus on the issue under discussion at the time. |
Sometimes you need access in the advice body to the actual value that was returned. You
can use the form of @AfterReturning
that binds the return value for this:
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.AfterReturning; @Aspect public class AfterReturningExample { @AfterReturning( pointcut="com.xyz.myapp.SystemArchitecture.dataAccessOperation()", returning="retVal") public void doAccessCheck(Object retVal) { // ... } }
The name used in the returning
attribute must correspond to the name of a parameter in
the advice method. When a method execution returns, the return value will be passed to
the advice method as the corresponding argument value. A returning
clause also
restricts matching to only those method executions that return a value of the specified
type ( Object
in this case, which will match any return value).
Please note that it is not possible to return a totally different reference when using after-returning advice.
After throwing advice runs when a matched method execution exits by throwing an
exception. It is declared using the @AfterThrowing
annotation:
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.AfterThrowing; @Aspect public class AfterThrowingExample { @AfterThrowing("com.xyz.myapp.SystemArchitecture.dataAccessOperation()") public void doRecoveryActions() { // ... } }
Often you want the advice to run only when exceptions of a given type are thrown, and
you also often need access to the thrown exception in the advice body. Use the
throwing
attribute to both restrict matching (if desired, use Throwable
as the
exception type otherwise) and bind the thrown exception to an advice parameter.
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.AfterThrowing; @Aspect public class AfterThrowingExample { @AfterThrowing( pointcut="com.xyz.myapp.SystemArchitecture.dataAccessOperation()", throwing="ex") public void doRecoveryActions(DataAccessException ex) { // ... } }
The name used in the throwing
attribute must correspond to the name of a parameter in
the advice method. When a method execution exits by throwing an exception, the exception
will be passed to the advice method as the corresponding argument value. A throwing
clause also restricts matching to only those method executions that throw an exception
of the specified type ( DataAccessException
in this case).
After (finally) advice runs however a matched method execution exits. It is declared
using the @After
annotation. After advice must be prepared to handle both normal and
exception return conditions. It is typically used for releasing resources, etc.
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.After; @Aspect public class AfterFinallyExample { @After("com.xyz.myapp.SystemArchitecture.dataAccessOperation()") public void doReleaseLock() { // ... } }
The final kind of advice is around advice. Around advice runs "around" a matched method execution. It has the opportunity to do work both before and after the method executes, and to determine when, how, and even if, the method actually gets to execute at all. Around advice is often used if you need to share state before and after a method execution in a thread-safe manner (starting and stopping a timer for example). Always use the least powerful form of advice that meets your requirements (i.e. don’t use around advice if simple before advice would do).
Around advice is declared using the @Around
annotation. The first parameter of the
advice method must be of type ProceedingJoinPoint
. Within the body of the advice,
calling proceed()
on the ProceedingJoinPoint
causes the underlying method to
execute. The proceed
method may also be called passing in an Object[]
- the values
in the array will be used as the arguments to the method execution when it proceeds.
Note | |
---|---|
The behavior of proceed when called with an Object[] is a little different than the behavior of proceed for around advice compiled by the AspectJ compiler. For around advice written using the traditional AspectJ language, the number of arguments passed to proceed must match the number of arguments passed to the around advice (not the number of arguments taken by the underlying join point), and the value passed to proceed in a given argument position supplants the original value at the join point for the entity the value was bound to (Don’t worry if this doesn’t make sense right now!). The approach taken by Spring is simpler and a better match to its proxy-based, execution only semantics. You only need to be aware of this difference if you are compiling @AspectJ aspects written for Spring and using proceed with arguments with the AspectJ compiler and weaver. There is a way to write such aspects that is 100% compatible across both Spring AOP and AspectJ, and this is discussed in the following section on advice parameters. |
import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.Around; import org.aspectj.lang.ProceedingJoinPoint; @Aspect public class AroundExample { @Around("com.xyz.myapp.SystemArchitecture.businessService()") public Object doBasicProfiling(ProceedingJoinPoint pjp) throws Throwable { // start stopwatch Object retVal = pjp.proceed(); // stop stopwatch return retVal; } }
The value returned by the around advice will be the return value seen by the caller of the method. A simple caching aspect for example could return a value from a cache if it has one, and invoke proceed() if it does not. Note that proceed may be invoked once, many times, or not at all within the body of the around advice, all of these are quite legal.
Spring offers fully typed advice - meaning that you declare the parameters you need
in the advice signature (as we saw for the returning and throwing examples above) rather
than work with Object[]
arrays all the time. We’ll see how to make argument and other
contextual values available to the advice body in a moment. First let’s take a look at
how to write generic advice that can find out about the method the advice is currently
advising.
Any advice method may declare as its first parameter, a parameter of type
org.aspectj.lang.JoinPoint
(please note that around advice is required to declare
a first parameter of type ProceedingJoinPoint
, which is a subclass of JoinPoint
. The
JoinPoint
interface provides a number of useful methods such as getArgs()
(returns
the method arguments), getThis()
(returns the proxy object), getTarget()
(returns
the target object), getSignature()
(returns a description of the method that is being
advised) and toString()
(prints a useful description of the method being advised).
Please do consult the javadocs for full details.
We’ve already seen how to bind the returned value or exception value (using after
returning and after throwing advice). To make argument values available to the advice
body, you can use the binding form of args
. If a parameter name is used in place of a
type name in an args expression, then the value of the corresponding argument will be
passed as the parameter value when the advice is invoked. An example should make this
clearer. Suppose you want to advise the execution of dao operations that take an Account
object as the first parameter, and you need access to the account in the advice body.
You could write the following:
@Before("com.xyz.myapp.SystemArchitecture.dataAccessOperation() && args(account,..)") public void validateAccount(Account account) { // ... }
The args(account,..)
part of the pointcut expression serves two purposes: firstly, it
restricts matching to only those method executions where the method takes at least one
parameter, and the argument passed to that parameter is an instance of Account
;
secondly, it makes the actual Account
object available to the advice via the account
parameter.
Another way of writing this is to declare a pointcut that "provides" the Account
object value when it matches a join point, and then just refer to the named pointcut
from the advice. This would look as follows:
@Pointcut("com.xyz.myapp.SystemArchitecture.dataAccessOperation() && args(account,..)") private void accountDataAccessOperation(Account account) {} @Before("accountDataAccessOperation(account)") public void validateAccount(Account account) { // ... }
The interested reader is once more referred to the AspectJ programming guide for more details.
The proxy object ( this
), target object ( target
), and annotations ( @within,
@target, @annotation, @args
) can all be bound in a similar fashion. The following
example shows how you could match the execution of methods annotated with an
@Auditable
annotation, and extract the audit code.
First the definition of the @Auditable
annotation:
@Retention(RetentionPolicy.RUNTIME) @Target(ElementType.METHOD) public @interface Auditable { AuditCode value(); }
And then the advice that matches the execution of @Auditable
methods:
@Before("com.xyz.lib.Pointcuts.anyPublicMethod() && @annotation(auditable)") public void audit(Auditable auditable) { AuditCode code = auditable.value(); // ... }
Spring AOP can handle generics used in class declarations and method parameters. Suppose you have a generic type like this:
public interface Sample<T> { void sampleGenericMethod(T param); void sampleGenericCollectionMethod(Collection>T> param); }
You can restrict interception of method types to certain parameter types by simply typing the advice parameter to the parameter type you want to intercept the method for:
@Before("execution(* ..Sample+.sampleGenericMethod(*)) && args(param)") public void beforeSampleMethod(MyType param) { // Advice implementation }
That this works is pretty obvious as we already discussed above. However, it’s worth pointing out that this won’t work for generic collections. So you cannot define a pointcut like this:
@Before("execution(* ..Sample+.sampleGenericCollectionMethod(*)) && args(param)") public void beforeSampleMethod(Collection<MyType> param) { // Advice implementation }
To make this work we would have to inspect every element of the collection, which is not
reasonable as we also cannot decide how to treat null
values in general. To achieve
something similar to this you have to type the parameter to Collection<?>
and manually
check the type of the elements.
The parameter binding in advice invocations relies on matching names used in pointcut expressions to declared parameter names in (advice and pointcut) method signatures. Parameter names are not available through Java reflection, so Spring AOP uses the following strategies to determine parameter names:
@Before(value="com.xyz.lib.Pointcuts.anyPublicMethod() && target(bean) && @annotation(auditable)", argNames="bean,auditable") public void audit(Object bean, Auditable auditable) { AuditCode code = auditable.value(); // ... use code and bean }
If the first parameter is of the JoinPoint
, ProceedingJoinPoint
, or
JoinPoint.StaticPart
type, you may leave out the name of the parameter from the value
of the "argNames" attribute. For example, if you modify the preceding advice to receive
the join point object, the "argNames" attribute need not include it:
@Before(value="com.xyz.lib.Pointcuts.anyPublicMethod() && target(bean) && @annotation(auditable)", argNames="bean,auditable") public void audit(JoinPoint jp, Object bean, Auditable auditable) { AuditCode code = auditable.value(); // ... use code, bean, and jp }
The special treatment given to the first parameter of the JoinPoint
,
ProceedingJoinPoint
, and JoinPoint.StaticPart
types is particularly convenient for
advice that do not collect any other join point context. In such situations, you may
simply omit the "argNames" attribute. For example, the following advice need not declare
the "argNames" attribute:
@Before("com.xyz.lib.Pointcuts.anyPublicMethod()") public void audit(JoinPoint jp) { // ... use jp }
'argNames'
attribute is a little clumsy, so if the 'argNames'
attribute
has not been specified, then Spring AOP will look at the debug information for the
class and try to determine the parameter names from the local variable table. This
information will be present as long as the classes have been compiled with debug
information ( '-g:vars'
at a minimum). The consequences of compiling with this flag
on are: (1) your code will be slightly easier to understand (reverse engineer), (2)
the class file sizes will be very slightly bigger (typically inconsequential), (3) the
optimization to remove unused local variables will not be applied by your compiler. In
other words, you should encounter no difficulties building with this flag on.
Note | |
---|---|
If an @AspectJ aspect has been compiled by the AspectJ compiler (ajc) even without the debug information then there is no need to add the argNames attribute as the compiler will retain the needed information. |
AmbiguousBindingException
will be
thrown.
IllegalArgumentException
will be thrown.
We remarked earlier that we would describe how to write a proceed call with arguments that works consistently across Spring AOP and AspectJ. The solution is simply to ensure that the advice signature binds each of the method parameters in order. For example:
@Around("execution(List<Account> find*(..)) && " + "com.xyz.myapp.SystemArchitecture.inDataAccessLayer() && " + "args(accountHolderNamePattern)") public Object preProcessQueryPattern(ProceedingJoinPoint pjp, String accountHolderNamePattern) throws Throwable { String newPattern = preProcess(accountHolderNamePattern); return pjp.proceed(new Object[] {newPattern}); }
In many cases you will be doing this binding anyway (as in the example above).
What happens when multiple pieces of advice all want to run at the same join point? Spring AOP follows the same precedence rules as AspectJ to determine the order of advice execution. The highest precedence advice runs first "on the way in" (so given two pieces of before advice, the one with highest precedence runs first). "On the way out" from a join point, the highest precedence advice runs last (so given two pieces of after advice, the one with the highest precedence will run second).
When two pieces of advice defined in different aspects both need to run at the same
join point, unless you specify otherwise the order of execution is undefined. You can
control the order of execution by specifying precedence. This is done in the normal
Spring way by either implementing the org.springframework.core.Ordered
interface in
the aspect class or annotating it with the Order
annotation. Given two aspects, the
aspect returning the lower value from Ordered.getValue()
(or the annotation value) has
the higher precedence.
When two pieces of advice defined in the same aspect both need to run at the same join point, the ordering is undefined (since there is no way to retrieve the declaration order via reflection for javac-compiled classes). Consider collapsing such advice methods into one advice method per join point in each aspect class, or refactor the pieces of advice into separate aspect classes - which can be ordered at the aspect level.
Introductions (known as inter-type declarations in AspectJ) enable an aspect to declare that advised objects implement a given interface, and to provide an implementation of that interface on behalf of those objects.
An introduction is made using the @DeclareParents
annotation. This annotation is used
to declare that matching types have a new parent (hence the name). For example, given an
interface UsageTracked
, and an implementation of that interface DefaultUsageTracked
,
the following aspect declares that all implementors of service interfaces also implement
the UsageTracked
interface. (In order to expose statistics via JMX for example.)
@Aspect public class UsageTracking { @DeclareParents(value="com.xzy.myapp.service.*+", defaultImpl=DefaultUsageTracked.class) public static UsageTracked mixin; @Before("com.xyz.myapp.SystemArchitecture.businessService() && this(usageTracked)") public void recordUsage(UsageTracked usageTracked) { usageTracked.incrementUseCount(); } }
The interface to be implemented is determined by the type of the annotated field. The
value
attribute of the @DeclareParents
annotation is an AspectJ type pattern :- any
bean of a matching type will implement the UsageTracked interface. Note that in the
before advice of the above example, service beans can be directly used as
implementations of the UsageTracked
interface. If accessing a bean programmatically
you would write the following:
UsageTracked usageTracked = (UsageTracked) context.getBean("myService");
Note | |
---|---|
(This is an advanced topic, so if you are just starting out with AOP you can safely skip it until later.) |
By default there will be a single instance of each aspect within the application
context. AspectJ calls this the singleton instantiation model. It is possible to define
aspects with alternate lifecycles :- Spring supports AspectJ’s perthis
and pertarget
instantiation models ( percflow, percflowbelow,
and pertypewithin
are not currently
supported).
A "perthis" aspect is declared by specifying a perthis
clause in the @Aspect
annotation. Let’s look at an example, and then we’ll explain how it works.
@Aspect("perthis(com.xyz.myapp.SystemArchitecture.businessService())") public class MyAspect { private int someState; @Before(com.xyz.myapp.SystemArchitecture.businessService()) public void recordServiceUsage() { // ... } }
The effect of the 'perthis'
clause is that one aspect instance will be created for
each unique service object executing a business service (each unique object bound to
this at join points matched by the pointcut expression). The aspect instance is
created the first time that a method is invoked on the service object. The aspect goes
out of scope when the service object goes out of scope. Before the aspect instance is
created, none of the advice within it executes. As soon as the aspect instance has been
created, the advice declared within it will execute at matched join points, but only
when the service object is the one this aspect is associated with. See the AspectJ
programming guide for more information on per-clauses.
The 'pertarget'
instantiation model works in exactly the same way as perthis, but
creates one aspect instance for each unique target object at matched join points.
Now that you have seen how all the constituent parts work, let’s put them together to do something useful!
The execution of business services can sometimes fail due to concurrency issues (for
example, deadlock loser). If the operation is retried, it is quite likely to succeed
next time round. For business services where it is appropriate to retry in such
conditions (idempotent operations that don’t need to go back to the user for conflict
resolution), we’d like to transparently retry the operation to avoid the client seeing a
PessimisticLockingFailureException
. This is a requirement that clearly cuts across
multiple services in the service layer, and hence is ideal for implementing via an
aspect.
Because we want to retry the operation, we will need to use around advice so that we can call proceed multiple times. Here’s how the basic aspect implementation looks:
@Aspect public class ConcurrentOperationExecutor implements Ordered { private static final int DEFAULT_MAX_RETRIES = 2; private int maxRetries = DEFAULT_MAX_RETRIES; private int order = 1; public void setMaxRetries(int maxRetries) { this.maxRetries = maxRetries; } public int getOrder() { return this.order; } public void setOrder(int order) { this.order = order; } @Around("com.xyz.myapp.SystemArchitecture.businessService()") public Object doConcurrentOperation(ProceedingJoinPoint pjp) throws Throwable { int numAttempts = 0; PessimisticLockingFailureException lockFailureException; do { numAttempts++; try { return pjp.proceed(); } catch(PessimisticLockingFailureException ex) { lockFailureException = ex; } } while(numAttempts <= this.maxRetries); throw lockFailureException; } }
Note that the aspect implements the Ordered
interface so we can set the precedence of
the aspect higher than the transaction advice (we want a fresh transaction each time we
retry). The maxRetries
and order
properties will both be configured by Spring. The
main action happens in the doConcurrentOperation
around advice. Notice that for the
moment we’re applying the retry logic to all businessService()s
. We try to proceed,
and if we fail with an PessimisticLockingFailureException
we simply try again unless
we have exhausted all of our retry attempts.
The corresponding Spring configuration is:
<aop:aspectj-autoproxy/> <bean id="concurrentOperationExecutor" class="com.xyz.myapp.service.impl.ConcurrentOperationExecutor"> <property name="maxRetries" value="3"/> <property name="order" value="100"/> </bean>
To refine the aspect so that it only retries idempotent operations, we might define an
Idempotent
annotation:
@Retention(RetentionPolicy.RUNTIME) public @interface Idempotent { // marker annotation }
and use the annotation to annotate the implementation of service operations. The change
to the aspect to only retry idempotent operations simply involves refining the pointcut
expression so that only @Idempotent
operations match:
@Around("com.xyz.myapp.SystemArchitecture.businessService() && " + "@annotation(com.xyz.myapp.service.Idempotent)") public Object doConcurrentOperation(ProceedingJoinPoint pjp) throws Throwable { ... }
If you prefer an XML-based format, then Spring also offers support for defining aspects using the new "aop" namespace tags. The exact same pointcut expressions and advice kinds are supported as when using the @AspectJ style, hence in this section we will focus on the new syntax and refer the reader to the discussion in the previous section (Section 9.2, “@AspectJ support”) for an understanding of writing pointcut expressions and the binding of advice parameters.
To use the aop namespace tags described in this section, you need to import the spring-aop schema as described in Chapter 34, XML Schema-based configuration. See Section 34.2.7, “the aop schema” for how to import the tags in the aop namespace.
Within your Spring configurations, all aspect and advisor elements must be placed within
an <aop:config>
element (you can have more than one <aop:config>
element in an
application context configuration). An <aop:config>
element can contain pointcut,
advisor, and aspect elements (note these must be declared in that order).
Warning | |
---|---|
The |
Using the schema support, an aspect is simply a regular Java object defined as a bean in your Spring application context. The state and behavior is captured in the fields and methods of the object, and the pointcut and advice information is captured in the XML.
An aspect is declared using the <aop:aspect> element, and the backing bean is referenced
using the ref
attribute:
<aop:config> <aop:aspect id="myAspect" ref="aBean"> ... </aop:aspect> </aop:config> <bean id="aBean" class="..."> ... </bean>
The bean backing the aspect (" aBean
" in this case) can of course be configured and
dependency injected just like any other Spring bean.
A named pointcut can be declared inside an <aop:config> element, enabling the pointcut definition to be shared across several aspects and advisors.
A pointcut representing the execution of any business service in the service layer could be defined as follows:
<aop:config> <aop:pointcut id="businessService" expression="execution(* com.xyz.myapp.service.*.*(..))"/> </aop:config>
Note that the pointcut expression itself is using the same AspectJ pointcut expression language as described in Section 9.2, “@AspectJ support”. If you are using the schema based declaration style, you can refer to named pointcuts defined in types (@Aspects) within the pointcut expression. Another way of defining the above pointcut would be:
<aop:config> <aop:pointcut id="businessService" expression="com.xyz.myapp.SystemArchitecture.businessService()"/> </aop:config>
Assuming you have a SystemArchitecture
aspect as described in the section called “Sharing common pointcut definitions”.
Declaring a pointcut inside an aspect is very similar to declaring a top-level pointcut:
<aop:config> <aop:aspect id="myAspect" ref="aBean"> <aop:pointcut id="businessService" expression="execution(* com.xyz.myapp.service.*.*(..))"/> ... </aop:aspect> </aop:config>
Much the same way in an @AspectJ aspect, pointcuts declared using the schema based definition style may collect join point context. For example, the following pointcut collects the this object as the join point context and passes it to advice:
<aop:config> <aop:aspect id="myAspect" ref="aBean"> <aop:pointcut id="businessService" expression="execution(* com.xyz.myapp.service.*.*(..)) && this(service)"/> <aop:before pointcut-ref="businessService" method="monitor"/> ... </aop:aspect> </aop:config>
The advice must be declared to receive the collected join point context by including parameters of the matching names:
public void monitor(Object service) { ... }
When combining pointcut sub-expressions, && is awkward within an XML document, and so the keywords and, or and not can be used in place of &&, || and ! respectively. For example, the previous pointcut may be better written as:
<aop:config> <aop:aspect id="myAspect" ref="aBean"> <aop:pointcut id="businessService" expression="execution(* com.xyz.myapp.service.*.*(..)) **and** this(service)"/> <aop:before pointcut-ref="businessService" method="monitor"/> ... </aop:aspect> </aop:config>
Note that pointcuts defined in this way are referred to by their XML id and cannot be used as named pointcuts to form composite pointcuts. The named pointcut support in the schema based definition style is thus more limited than that offered by the @AspectJ style.
The same five advice kinds are supported as for the @AspectJ style, and they have exactly the same semantics.
Before advice runs before a matched method execution. It is declared inside an
<aop:aspect>
using the <aop:before> element.
<aop:aspect id="beforeExample" ref="aBean"> <aop:before pointcut-ref="dataAccessOperation" method="doAccessCheck"/> ... </aop:aspect>
Here dataAccessOperation
is the id of a pointcut defined at the top ( <aop:config>
)
level. To define the pointcut inline instead, replace the pointcut-ref
attribute with
a pointcut
attribute:
<aop:aspect id="beforeExample" ref="aBean"> <aop:before pointcut="execution(* com.xyz.myapp.dao.*.*(..))" method="doAccessCheck"/> ... </aop:aspect>
As we noted in the discussion of the @AspectJ style, using named pointcuts can significantly improve the readability of your code.
The method attribute identifies a method ( doAccessCheck
) that provides the body of
the advice. This method must be defined for the bean referenced by the aspect element
containing the advice. Before a data access operation is executed (a method execution
join point matched by the pointcut expression), the "doAccessCheck" method on the aspect
bean will be invoked.
After returning advice runs when a matched method execution completes normally. It is
declared inside an <aop:aspect>
in the same way as before advice. For example:
<aop:aspect id="afterReturningExample" ref="aBean"> <aop:after-returning pointcut-ref="dataAccessOperation" method="doAccessCheck"/> ... </aop:aspect>
Just as in the @AspectJ style, it is possible to get hold of the return value within the advice body. Use the returning attribute to specify the name of the parameter to which the return value should be passed:
<aop:aspect id="afterReturningExample" ref="aBean"> <aop:after-returning pointcut-ref="dataAccessOperation" returning="retVal" method="doAccessCheck"/> ... </aop:aspect>
The doAccessCheck method must declare a parameter named retVal
. The type of this
parameter constrains matching in the same way as described for @AfterReturning. For
example, the method signature may be declared as:
public void doAccessCheck(Object retVal) {...
After throwing advice executes when a matched method execution exits by throwing an
exception. It is declared inside an <aop:aspect>
using the after-throwing element:
<aop:aspect id="afterThrowingExample" ref="aBean"> <aop:after-throwing pointcut-ref="dataAccessOperation" method="doRecoveryActions"/> ... </aop:aspect>
Just as in the @AspectJ style, it is possible to get hold of the thrown exception within the advice body. Use the throwing attribute to specify the name of the parameter to which the exception should be passed:
<aop:aspect id="afterThrowingExample" ref="aBean"> <aop:after-throwing pointcut-ref="dataAccessOperation" throwing="dataAccessEx" method="doRecoveryActions"/> ... </aop:aspect>
The doRecoveryActions method must declare a parameter named dataAccessEx
. The type of
this parameter constrains matching in the same way as described for @AfterThrowing. For
example, the method signature may be declared as:
public void doRecoveryActions(DataAccessException dataAccessEx) {...
After (finally) advice runs however a matched method execution exits. It is declared
using the after
element:
<aop:aspect id="afterFinallyExample" ref="aBean"> <aop:after pointcut-ref="dataAccessOperation" method="doReleaseLock"/> ... </aop:aspect>
The final kind of advice is around advice. Around advice runs "around" a matched method execution. It has the opportunity to do work both before and after the method executes, and to determine when, how, and even if, the method actually gets to execute at all. Around advice is often used if you need to share state before and after a method execution in a thread-safe manner (starting and stopping a timer for example). Always use the least powerful form of advice that meets your requirements; don’t use around advice if simple before advice would do.
Around advice is declared using the aop:around
element. The first parameter of the
advice method must be of type ProceedingJoinPoint
. Within the body of the advice,
calling proceed()
on the ProceedingJoinPoint
causes the underlying method to
execute. The proceed
method may also be calling passing in an Object[]
- the values
in the array will be used as the arguments to the method execution when it proceeds. See
the section called “Around advice” for notes on calling proceed with an Object[]
.
<aop:aspect id="aroundExample" ref="aBean"> <aop:around pointcut-ref="businessService" method="doBasicProfiling"/> ... </aop:aspect>
The implementation of the doBasicProfiling
advice would be exactly the same as in the
@AspectJ example (minus the annotation of course):
public Object doBasicProfiling(ProceedingJoinPoint pjp) throws Throwable { // start stopwatch Object retVal = pjp.proceed(); // stop stopwatch return retVal; }
The schema based declaration style supports fully typed advice in the same way as
described for the @AspectJ support - by matching pointcut parameters by name against
advice method parameters. See the section called “Advice parameters” for details. If you wish
to explicitly specify argument names for the advice methods (not relying on the
detection strategies previously described) then this is done using the arg-names
attribute of the advice element, which is treated in the same manner to the "argNames"
attribute in an advice annotation as described in the section called “Determining argument names”.
For example:
<aop:before pointcut="com.xyz.lib.Pointcuts.anyPublicMethod() and @annotation(auditable)" method="audit" arg-names="auditable"/>
The arg-names
attribute accepts a comma-delimited list of parameter names.
Find below a slightly more involved example of the XSD-based approach that illustrates some around advice used in conjunction with a number of strongly typed parameters.
package x.y.service; public interface FooService { Foo getFoo(String fooName, int age); } public class DefaultFooService implements FooService { public Foo getFoo(String name, int age) { return new Foo(name, age); } }
Next up is the aspect. Notice the fact that the profile(..)
method accepts a number of
strongly-typed parameters, the first of which happens to be the join point used to
proceed with the method call: the presence of this parameter is an indication that the
profile(..)
is to be used as around
advice:
package x.y; import org.aspectj.lang.ProceedingJoinPoint; import org.springframework.util.StopWatch; public class SimpleProfiler { public Object profile(ProceedingJoinPoint call, String name, int age) throws Throwable { StopWatch clock = new StopWatch("Profiling for '" + name + "' and '" + age + "'"); try { clock.start(call.toShortString()); return call.proceed(); } finally { clock.stop(); System.out.println(clock.prettyPrint()); } } }
Finally, here is the XML configuration that is required to effect the execution of the above advice for a particular join point:
<beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:aop="http://www.springframework.org/schema/aop" xsi:schemaLocation=" http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd http://www.springframework.org/schema/aop http://www.springframework.org/schema/aop/spring-aop.xsd"> <!-- this is the object that will be proxied by Spring's AOP infrastructure --> <bean id="fooService" class="x.y.service.DefaultFooService"/> <!-- this is the actual advice itself --> <bean id="profiler" class="x.y.SimpleProfiler"/> <aop:config> <aop:aspect ref="profiler"> <aop:pointcut id="theExecutionOfSomeFooServiceMethod" expression="execution(* x.y.service.FooService.getFoo(String,int)) and args(name, age)"/> <aop:around pointcut-ref="theExecutionOfSomeFooServiceMethod" method="profile"/> </aop:aspect> </aop:config> </beans>
If we had the following driver script, we would get output something like this on standard output:
import org.springframework.beans.factory.BeanFactory; import org.springframework.context.support.ClassPathXmlApplicationContext; import x.y.service.FooService; public final class Boot { public static void main(final String[] args) throws Exception { BeanFactory ctx = new ClassPathXmlApplicationContext("x/y/plain.xml"); FooService foo = (FooService) ctx.getBean("fooService"); foo.getFoo("Pengo", 12); } }
StopWatch Profiling for 'Pengo and 12': running time (millis) = 0 ----------------------------------------- ms % Task name ----------------------------------------- 00000 ? execution(getFoo)
When multiple advice needs to execute at the same join point (executing method) the
ordering rules are as described in the section called “Advice ordering”. The precedence
between aspects is determined by either adding the Order
annotation to the bean
backing the aspect or by having the bean implement the Ordered
interface.
Introductions (known as inter-type declarations in AspectJ) enable an aspect to declare that advised objects implement a given interface, and to provide an implementation of that interface on behalf of those objects.
An introduction is made using the aop:declare-parents
element inside an aop:aspect
This element is used to declare that matching types have a new parent (hence the name).
For example, given an interface UsageTracked
, and an implementation of that interface
DefaultUsageTracked
, the following aspect declares that all implementors of service
interfaces also implement the UsageTracked
interface. (In order to expose statistics
via JMX for example.)
<aop:aspect id="usageTrackerAspect" ref="usageTracking"> <aop:declare-parents types-matching="com.xzy.myapp.service.*+" implement-interface="com.xyz.myapp.service.tracking.UsageTracked" default-impl="com.xyz.myapp.service.tracking.DefaultUsageTracked"/> <aop:before pointcut="com.xyz.myapp.SystemArchitecture.businessService() and this(usageTracked)" method="recordUsage"/> </aop:aspect>
The class backing the usageTracking
bean would contain the method:
public void recordUsage(UsageTracked usageTracked) { usageTracked.incrementUseCount(); }
The interface to be implemented is determined by implement-interface
attribute. The
value of the types-matching
attribute is an AspectJ type pattern :- any bean of a
matching type will implement the UsageTracked
interface. Note that in the before
advice of the above example, service beans can be directly used as implementations of
the UsageTracked
interface. If accessing a bean programmatically you would write the
following:
UsageTracked usageTracked = (UsageTracked) context.getBean("myService");
The only supported instantiation model for schema-defined aspects is the singleton model. Other instantiation models may be supported in future releases.
The concept of "advisors" is brought forward from the AOP support defined in Spring 1.2 and does not have a direct equivalent in AspectJ. An advisor is like a small self-contained aspect that has a single piece of advice. The advice itself is represented by a bean, and must implement one of the advice interfaces described in Section 10.3.2, “Advice types in Spring”. Advisors can take advantage of AspectJ pointcut expressions though.
Spring supports the advisor concept with the <aop:advisor>
element. You will most
commonly see it used in conjunction with transactional advice, which also has its own
namespace support in Spring. Here’s how it looks:
<aop:config> <aop:pointcut id="businessService" expression="execution(* com.xyz.myapp.service.*.*(..))"/> <aop:advisor pointcut-ref="businessService" advice-ref="tx-advice"/> </aop:config> <tx:advice id="tx-advice"> <tx:attributes> <tx:method name="*" propagation="REQUIRED"/> </tx:attributes> </tx:advice>
As well as the pointcut-ref
attribute used in the above example, you can also use the
pointcut
attribute to define a pointcut expression inline.
To define the precedence of an advisor so that the advice can participate in ordering,
use the order
attribute to define the Ordered
value of the advisor.
Let’s see how the concurrent locking failure retry example from Section 9.2.7, “Example” looks when rewritten using the schema support.
The execution of business services can sometimes fail due to concurrency issues (for
example, deadlock loser). If the operation is retried, it is quite likely it will
succeed next time round. For business services where it is appropriate to retry in such
conditions (idempotent operations that don’t need to go back to the user for conflict
resolution), we’d like to transparently retry the operation to avoid the client seeing a
PessimisticLockingFailureException
. This is a requirement that clearly cuts across
multiple services in the service layer, and hence is ideal for implementing via an
aspect.
Because we want to retry the operation, we’ll need to use around advice so that we can call proceed multiple times. Here’s how the basic aspect implementation looks (it’s just a regular Java class using the schema support):
public class ConcurrentOperationExecutor implements Ordered { private static final int DEFAULT_MAX_RETRIES = 2; private int maxRetries = DEFAULT_MAX_RETRIES; private int order = 1; public void setMaxRetries(int maxRetries) { this.maxRetries = maxRetries; } public int getOrder() { return this.order; } public void setOrder(int order) { this.order = order; } public Object doConcurrentOperation(ProceedingJoinPoint pjp) throws Throwable { int numAttempts = 0; PessimisticLockingFailureException lockFailureException; do { numAttempts++; try { return pjp.proceed(); } catch(PessimisticLockingFailureException ex) { lockFailureException = ex; } } while(numAttempts <= this.maxRetries); throw lockFailureException; } }
Note that the aspect implements the Ordered
interface so we can set the precedence of
the aspect higher than the transaction advice (we want a fresh transaction each time we
retry). The maxRetries
and order
properties will both be configured by Spring. The
main action happens in the doConcurrentOperation
around advice method. We try to
proceed, and if we fail with a PessimisticLockingFailureException
we simply try again
unless we have exhausted all of our retry attempts.
Note | |
---|---|
This class is identical to the one used in the @AspectJ example, but with the annotations removed. |
The corresponding Spring configuration is:
<aop:config> <aop:aspect id="concurrentOperationRetry" ref="concurrentOperationExecutor"> <aop:pointcut id="idempotentOperation" expression="execution(* com.xyz.myapp.service.*.*(..))"/> <aop:around pointcut-ref="idempotentOperation" method="doConcurrentOperation"/> </aop:aspect> </aop:config> <bean id="concurrentOperationExecutor" class="com.xyz.myapp.service.impl.ConcurrentOperationExecutor"> <property name="maxRetries" value="3"/> <property name="order" value="100"/> </bean>
Notice that for the time being we assume that all business services are idempotent. If
this is not the case we can refine the aspect so that it only retries genuinely
idempotent operations, by introducing an Idempotent
annotation:
@Retention(RetentionPolicy.RUNTIME) public @interface Idempotent { // marker annotation }
and using the annotation to annotate the implementation of service operations. The
change to the aspect to retry only idempotent operations simply involves refining the
pointcut expression so that only @Idempotent
operations match:
<aop:pointcut id="idempotentOperation" expression="execution(* com.xyz.myapp.service.*.*(..)) and @annotation(com.xyz.myapp.service.Idempotent)"/>
Once you have decided that an aspect is the best approach for implementing a given requirement, how do you decide between using Spring AOP or AspectJ, and between the Aspect language (code) style, @AspectJ annotation style, or the Spring XML style? These decisions are influenced by a number of factors including application requirements, development tools, and team familiarity with AOP.
Use the simplest thing that can work. Spring AOP is simpler than using full AspectJ as there is no requirement to introduce the AspectJ compiler / weaver into your development and build processes. If you only need to advise the execution of operations on Spring beans, then Spring AOP is the right choice. If you need to advise objects not managed by the Spring container (such as domain objects typically), then you will need to use AspectJ. You will also need to use AspectJ if you wish to advise join points other than simple method executions (for example, field get or set join points, and so on).
When using AspectJ, you have the choice of the AspectJ language syntax (also known as the "code style") or the @AspectJ annotation style. Clearly, if you are not using Java 5+ then the choice has been made for you… use the code style. If aspects play a large role in your design, and you are able to use the AspectJ Development Tools (AJDT) plugin for Eclipse, then the AspectJ language syntax is the preferred option: it is cleaner and simpler because the language was purposefully designed for writing aspects. If you are not using Eclipse, or have only a few aspects that do not play a major role in your application, then you may want to consider using the @AspectJ style and sticking with a regular Java compilation in your IDE, and adding an aspect weaving phase to your build script.
If you have chosen to use Spring AOP, then you have a choice of @AspectJ or XML style. There are various tradeoffs to consider.
The XML style will be most familiar to existing Spring users and it is backed by genuine POJOs. When using AOP as a tool to configure enterprise services then XML can be a good choice (a good test is whether you consider the pointcut expression to be a part of your configuration you might want to change independently). With the XML style arguably it is clearer from your configuration what aspects are present in the system.
The XML style has two disadvantages. Firstly it does not fully encapsulate the implementation of the requirement it addresses in a single place. The DRY principle says that there should be a single, unambiguous, authoritative representation of any piece of knowledge within a system. When using the XML style, the knowledge of how a requirement is implemented is split across the declaration of the backing bean class, and the XML in the configuration file. When using the @AspectJ style there is a single module - the aspect - in which this information is encapsulated. Secondly, the XML style is slightly more limited in what it can express than the @AspectJ style: only the "singleton" aspect instantiation model is supported, and it is not possible to combine named pointcuts declared in XML. For example, in the @AspectJ style you can write something like:
@Pointcut(execution(* get*())) public void propertyAccess() {} @Pointcut(execution(org.xyz.Account+ *(..)) public void operationReturningAnAccount() {} @Pointcut(propertyAccess() && operationReturningAnAccount()) public void accountPropertyAccess() {}
In the XML style I can declare the first two pointcuts:
<aop:pointcut id="propertyAccess" expression="execution(* get*())"/> <aop:pointcut id="operationReturningAnAccount" expression="execution(org.xyz.Account+ *(..))"/>
The downside of the XML approach is that you cannot define the
accountPropertyAccess
pointcut by combining these definitions.
The @AspectJ style supports additional instantiation models, and richer pointcut composition. It has the advantage of keeping the aspect as a modular unit. It also has the advantage the @AspectJ aspects can be understood (and thus consumed) both by Spring AOP and by AspectJ - so if you later decide you need the capabilities of AspectJ to implement additional requirements then it is very easy to migrate to an AspectJ-based approach. On balance the Spring team prefer the @AspectJ style whenever you have aspects that do more than simple "configuration" of enterprise services.
It is perfectly possible to mix @AspectJ style aspects using the autoproxying support,
schema-defined <aop:aspect>
aspects, <aop:advisor>
declared advisors and even
proxies and interceptors defined using the Spring 1.2 style in the same configuration.
All of these are implemented using the same underlying support mechanism and will
co-exist without any difficulty.
Spring AOP uses either JDK dynamic proxies or CGLIB to create the proxy for a given target object. (JDK dynamic proxies are preferred whenever you have a choice).
If the target object to be proxied implements at least one interface then a JDK dynamic proxy will be used. All of the interfaces implemented by the target type will be proxied. If the target object does not implement any interfaces then a CGLIB proxy will be created.
If you want to force the use of CGLIB proxying (for example, to proxy every method defined for the target object, not just those implemented by its interfaces) you can do so. However, there are some issues to consider:
final
methods cannot be advised, as they cannot be overridden.
To force the use of CGLIB proxies set the value of the proxy-target-class
attribute of
the <aop:config>
element to true:
<aop:config proxy-target-class="true"> <!-- other beans defined here... --> </aop:config>
To force CGLIB proxying when using the @AspectJ autoproxy support, set the
'proxy-target-class'
attribute of the <aop:aspectj-autoproxy>
element to true
:
<aop:aspectj-autoproxy proxy-target-class="true"/>
Note | |
---|---|
Multiple To be clear: using |
Spring AOP is proxy-based. It is vitally important that you grasp the semantics of what that last statement actually means before you write your own aspects or use any of the Spring AOP-based aspects supplied with the Spring Framework.
Consider first the scenario where you have a plain-vanilla, un-proxied, nothing-special-about-it, straight object reference, as illustrated by the following code snippet.
public class SimplePojo implements Pojo { public void foo() { // this next method invocation is a direct call on the this reference this.bar(); } public void bar() { // some logic... } }
If you invoke a method on an object reference, the method is invoked directly on that object reference, as can be seen below.
public class Main { public static void main(String[] args) { Pojo pojo = new SimplePojo(); // this is a direct method call on the pojo reference pojo.foo(); } }
Things change slightly when the reference that client code has is a proxy. Consider the following diagram and code snippet.
public class Main { public static void main(String[] args) { ProxyFactory factory = new ProxyFactory(new SimplePojo()); factory.addInterface(Pojo.class); factory.addAdvice(new RetryAdvice()); Pojo pojo = (Pojo) factory.getProxy(); // this is a method call on the proxy! pojo.foo(); } }
The key thing to understand here is that the client code inside the main(..)
of the
Main
class has a reference to the proxy. This means that method calls on that
object reference will be calls on the proxy, and as such the proxy will be able to
delegate to all of the interceptors (advice) that are relevant to that particular method
call. However, once the call has finally reached the target object, the SimplePojo
reference in this case, any method calls that it may make on itself, such as
this.bar()
or this.foo()
, are going to be invoked against the this reference,
and not the proxy. This has important implications. It means that self-invocation is
not going to result in the advice associated with a method invocation getting a
chance to execute.
Okay, so what is to be done about this? The best approach (the term best is used loosely here) is to refactor your code such that the self-invocation does not happen. For sure, this does entail some work on your part, but it is the best, least-invasive approach. The next approach is absolutely horrendous, and I am almost reticent to point it out precisely because it is so horrendous. You can (choke!) totally tie the logic within your class to Spring AOP by doing this:
public class SimplePojo implements Pojo { public void foo() { // this works, but... gah! ((Pojo) AopContext.currentProxy()).bar(); } public void bar() { // some logic... } }
This totally couples your code to Spring AOP, and it makes the class itself aware of the fact that it is being used in an AOP context, which flies in the face of AOP. It also requires some additional configuration when the proxy is being created:
public class Main { public static void main(String[] args) { ProxyFactory factory = new ProxyFactory(new SimplePojo()); factory.adddInterface(Pojo.class); factory.addAdvice(new RetryAdvice()); factory.setExposeProxy(true); Pojo pojo = (Pojo) factory.getProxy(); // this is a method call on the proxy! pojo.foo(); } }
Finally, it must be noted that AspectJ does not have this self-invocation issue because it is not a proxy-based AOP framework.
In addition to declaring aspects in your configuration using either <aop:config>
or
<aop:aspectj-autoproxy>
, it is also possible programmatically to create proxies that
advise target objects. For the full details of Spring’s AOP API, see the next chapter.
Here we want to focus on the ability to automatically create proxies using @AspectJ
aspects.
The class org.springframework.aop.aspectj.annotation.AspectJProxyFactory
can be used
to create a proxy for a target object that is advised by one or more @AspectJ aspects.
Basic usage for this class is very simple, as illustrated below. See the javadocs for
full information.
// create a factory that can generate a proxy for the given target object AspectJProxyFactory factory = new AspectJProxyFactory(targetObject); // add an aspect, the class must be an @AspectJ aspect // you can call this as many times as you need with different aspects factory.addAspect(SecurityManager.class); // you can also add existing aspect instances, the type of the object supplied must be an @AspectJ aspect factory.addAspect(usageTracker); // now get the proxy object... MyInterfaceType proxy = factory.getProxy();
Everything we’ve covered so far in this chapter is pure Spring AOP. In this section, we’re going to look at how you can use the AspectJ compiler/weaver instead of, or in addition to, Spring AOP if your needs go beyond the facilities offered by Spring AOP alone.
Spring ships with a small AspectJ aspect library, which is available standalone in your
distribution as spring-aspects.jar
; you’ll need to add this to your classpath in order
to use the aspects in it. Section 9.8.1, “Using AspectJ to dependency inject domain objects with Spring” and Section 9.8.2, “Other Spring aspects for AspectJ” discuss the
content of this library and how you can use it. Section 9.8.3, “Configuring AspectJ aspects using Spring IoC” discusses how to
dependency inject AspectJ aspects that are woven using the AspectJ compiler. Finally,
Section 9.8.4, “Load-time weaving with AspectJ in the Spring Framework” provides an introduction to load-time weaving for Spring applications
using AspectJ.
The Spring container instantiates and configures beans defined in your application
context. It is also possible to ask a bean factory to configure a pre-existing
object given the name of a bean definition containing the configuration to be applied.
The spring-aspects.jar
contains an annotation-driven aspect that exploits this
capability to allow dependency injection of any object. The support is intended to
be used for objects created outside of the control of any container. Domain objects
often fall into this category because they are often created programmatically using the
new
operator, or by an ORM tool as a result of a database query.
The @Configurable
annotation marks a class as eligible for Spring-driven
configuration. In the simplest case it can be used just as a marker annotation:
package com.xyz.myapp.domain; import org.springframework.beans.factory.annotation.Configurable; @Configurable public class Account { // ... }
When used as a marker interface in this way, Spring will configure new instances of the
annotated type ( Account
in this case) using a bean definition (typically
prototype-scoped) with the same name as the fully-qualified type name (
com.xyz.myapp.domain.Account
). Since the default name for a bean is the
fully-qualified name of its type, a convenient way to declare the prototype definition
is simply to omit the id
attribute:
<bean class="com.xyz.myapp.domain.Account" scope="prototype"> <property name="fundsTransferService" ref="fundsTransferService"/> </bean>
If you want to explicitly specify the name of the prototype bean definition to use, you can do so directly in the annotation:
package com.xyz.myapp.domain; import org.springframework.beans.factory.annotation.Configurable; @Configurable("account") public class Account { // ... }
Spring will now look for a bean definition named " account
" and use that as the
definition to configure new Account
instances.
You can also use autowiring to avoid having to specify a dedicated bean definition at
all. To have Spring apply autowiring use the autowire
property of the
@Configurable
annotation: specify either @Configurable(autowire=Autowire.BY_TYPE)
or
@Configurable(autowire=Autowire.BY_NAME
for autowiring by type or by name
respectively. As an alternative, as of Spring 2.5 it is preferable to specify explicit,
annotation-driven dependency injection for your @Configurable
beans by using
@Autowired
or @Inject
at the field or method level (see Section 5.9, “Annotation-based container configuration”
for further details).
Finally you can enable Spring dependency checking for the object references in the newly
created and configured object by using the dependencyCheck
attribute (for example:
@Configurable(autowire=Autowire.BY_NAME,dependencyCheck=true)
). If this attribute is
set to true, then Spring will validate after configuration that all properties (which
are not primitives or collections) have been set.
Using the annotation on its own does nothing of course. It is the
AnnotationBeanConfigurerAspect
in spring-aspects.jar
that acts on the presence of
the annotation. In essence the aspect says "after returning from the initialization of a
new object of a type annotated with @Configurable
, configure the newly created object
using Spring in accordance with the properties of the annotation". In this context,
initialization refers to newly instantiated objects (e.g., objects instantiated with
the new
operator) as well as to Serializable
objects that are undergoing
deserialization (e.g., via
readResolve()).
Note | |
---|---|
One of the key phrases in the above paragraph is in essence. For most cases, the
exact semantics of after returning from the initialization of a new object will be
fine… in this context, after initialization means that the dependencies will be
injected after the object has been constructed - this means that the dependencies
will not be available for use in the constructor bodies of the class. If you want the
dependencies to be injected before the constructor bodies execute, and thus be
available for use in the body of the constructors, then you need to define this on the
@Configurable(preConstruction=true) You can find out more information about the language semantics of the various pointcut types in AspectJ in this appendix of the AspectJ Programming Guide. |
For this to work the annotated types must be woven with the AspectJ weaver - you can
either use a build-time Ant or Maven task to do this (see for example the
AspectJ Development
Environment Guide) or load-time weaving (see Section 9.8.4, “Load-time weaving with AspectJ in the Spring Framework”). The
AnnotationBeanConfigurerAspect
itself needs configuring by Spring (in order to obtain
a reference to the bean factory that is to be used to configure new objects). If you are
using Java based configuration simply add @EnableSpringConfigured
to any
@Configuration
class.
@Configuration @EnableSpringConfigured public class AppConfig { }
If you prefer XML based configuration, the Spring context
namespace defines a convenient context:spring-configured
element:
<context:spring-configured/>
Instances of @Configurable
objects created before the aspect has been configured
will result in a message being issued to the debug log and no configuration of the
object taking place. An example might be a bean in the Spring configuration that creates
domain objects when it is initialized by Spring. In this case you can use the
"depends-on" bean attribute to manually specify that the bean depends on the
configuration aspect.
<bean id="myService" class="com.xzy.myapp.service.MyService" depends-on="org.springframework.beans.factory.aspectj.AnnotationBeanConfigurerAspect"> <!-- ... --> </bean>
Note | |
---|---|
Do not activate |
One of the goals of the @Configurable
support is to enable independent unit testing of
domain objects without the difficulties associated with hard-coded lookups. If
@Configurable
types have not been woven by AspectJ then the annotation has no affect
during unit testing, and you can simply set mock or stub property references in the
object under test and proceed as normal. If @Configurable
types have been woven by
AspectJ then you can still unit test outside of the container as normal, but you will
see a warning message each time that you construct an @Configurable
object indicating
that it has not been configured by Spring.
The AnnotationBeanConfigurerAspect
used to implement the @Configurable
support is an
AspectJ singleton aspect. The scope of a singleton aspect is the same as the scope of
static
members, that is to say there is one aspect instance per classloader that
defines the type. This means that if you define multiple application contexts within the
same classloader hierarchy you need to consider where to define the
@EnableSpringConfigured
bean and where to place spring-aspects.jar
on the classpath.
Consider a typical Spring web-app configuration with a shared parent application context
defining common business services and everything needed to support them, and one child
application context per servlet containing definitions particular to that servlet. All
of these contexts will co-exist within the same classloader hierarchy, and so the
AnnotationBeanConfigurerAspect
can only hold a reference to one of them. In this case
we recommend defining the @EnableSpringConfigured
bean in the shared (parent)
application context: this defines the services that you are likely to want to inject
into domain objects. A consequence is that you cannot configure domain objects with
references to beans defined in the child (servlet-specific) contexts using the
@Configurable mechanism (probably not something you want to do anyway!).
When deploying multiple web-apps within the same container, ensure that each
web-application loads the types in spring-aspects.jar
using its own classloader (for
example, by placing spring-aspects.jar
in 'WEB-INF/lib'
). If spring-aspects.jar
is
only added to the container wide classpath (and hence loaded by the shared parent
classloader), all web applications will share the same aspect instance which is probably
not what you want.
In addition to the @Configurable
aspect, spring-aspects.jar
contains an AspectJ
aspect that can be used to drive Spring’s transaction management for types and methods
annotated with the @Transactional
annotation. This is primarily intended for users who
want to use the Spring Framework’s transaction support outside of the Spring container.
The aspect that interprets @Transactional
annotations is the
AnnotationTransactionAspect
. When using this aspect, you must annotate the
implementation class (and/or methods within that class), not the interface (if
any) that the class implements. AspectJ follows Java’s rule that annotations on
interfaces are not inherited.
A @Transactional
annotation on a class specifies the default transaction semantics for
the execution of any public operation in the class.
A @Transactional
annotation on a method within the class overrides the default
transaction semantics given by the class annotation (if present). Methods with public
,
protected
, and default visibility may all be annotated. Annotating protected
and
default visibility methods directly is the only way to get transaction demarcation for
the execution of such methods.
For AspectJ programmers that want to use the Spring configuration and transaction
management support but don’t want to (or cannot) use annotations, spring-aspects.jar
also contains abstract
aspects you can extend to provide your own pointcut
definitions. See the sources for the AbstractBeanConfigurerAspect
and
AbstractTransactionAspect
aspects for more information. As an example, the following
excerpt shows how you could write an aspect to configure all instances of objects
defined in the domain model using prototype bean definitions that match the
fully-qualified class names:
public aspect DomainObjectConfiguration extends AbstractBeanConfigurerAspect { public DomainObjectConfiguration() { setBeanWiringInfoResolver(new ClassNameBeanWiringInfoResolver()); } // the creation of a new bean (any object in the domain model) protected pointcut beanCreation(Object beanInstance) : initialization(new(..)) && SystemArchitecture.inDomainModel() && this(beanInstance); }
When using AspectJ aspects with Spring applications, it is natural to both want and
expect to be able to configure such aspects using Spring. The AspectJ runtime itself is
responsible for aspect creation, and the means of configuring the AspectJ created
aspects via Spring depends on the AspectJ instantiation model (the per-xxx
clause)
used by the aspect.
The majority of AspectJ aspects are singleton aspects. Configuration of these
aspects is very easy: simply create a bean definition referencing the aspect type as
normal, and include the bean attribute 'factory-method="aspectOf"'
. This ensures that
Spring obtains the aspect instance by asking AspectJ for it rather than trying to create
an instance itself. For example:
<bean id="profiler" class="com.xyz.profiler.Profiler" factory-method="aspectOf"> <property name="profilingStrategy" ref="jamonProfilingStrategy"/> </bean>
Non-singleton aspects are harder to configure: however it is possible to do so by
creating prototype bean definitions and using the @Configurable
support from
spring-aspects.jar
to configure the aspect instances once they have bean created by
the AspectJ runtime.
If you have some @AspectJ aspects that you want to weave with AspectJ (for example,
using load-time weaving for domain model types) and other @AspectJ aspects that you want
to use with Spring AOP, and these aspects are all configured using Spring, then you will
need to tell the Spring AOP @AspectJ autoproxying support which exact subset of the
@AspectJ aspects defined in the configuration should be used for autoproxying. You can
do this by using one or more <include/>
elements inside the <aop:aspectj-autoproxy/>
declaration. Each <include/>
element specifies a name pattern, and only beans with
names matched by at least one of the patterns will be used for Spring AOP autoproxy
configuration:
<aop:aspectj-autoproxy> <aop:include name="thisBean"/> <aop:include name="thatBean"/> </aop:aspectj-autoproxy>
Note | |
---|---|
Do not be misled by the name of the |
Load-time weaving (LTW) refers to the process of weaving AspectJ aspects into an application’s class files as they are being loaded into the Java virtual machine (JVM). The focus of this section is on configuring and using LTW in the specific context of the Spring Framework: this section is not an introduction to LTW though. For full details on the specifics of LTW and configuring LTW with just AspectJ (with Spring not being involved at all), see the LTW section of the AspectJ Development Environment Guide.
The value-add that the Spring Framework brings to AspectJ LTW is in enabling much finer-grained control over the weaving process. Vanilla AspectJ LTW is effected using a Java (5+) agent, which is switched on by specifying a VM argument when starting up a JVM. It is thus a JVM-wide setting, which may be fine in some situations, but often is a little too coarse. Spring-enabled LTW enables you to switch on LTW on a per-ClassLoader basis, which obviously is more fine-grained and which can make more sense in a single-JVM-multiple-application environment (such as is found in a typical application server environment).
Further, in certain environments, this support enables
load-time weaving without making any modifications to the application server’s launch
script that will be needed to add -javaagent:path/to/aspectjweaver.jar
or (as we
describe later in this section)
-javaagent:path/to/org.springframework.instrument-{version}.jar
(previously named
spring-agent.jar
). Developers simply modify one or more files that form the
application context to enable load-time weaving instead of relying on administrators who
typically are in charge of the deployment configuration such as the launch script.
Now that the sales pitch is over, let us first walk through a quick example of AspectJ LTW using Spring, followed by detailed specifics about elements introduced in the following example. For a complete example, please see the Petclinic sample application.
Let us assume that you are an application developer who has been tasked with diagnosing the cause of some performance problems in a system. Rather than break out a profiling tool, what we are going to do is switch on a simple profiling aspect that will enable us to very quickly get some performance metrics, so that we can then apply a finer-grained profiling tool to that specific area immediately afterwards.
Note | |
---|---|
The example presented here uses XML style configuration, it is also possible to
configure and use @AspectJ with Java Configuration. Specifically the
|
Here is the profiling aspect. Nothing too fancy, just a quick-and-dirty time-based profiler, using the @AspectJ-style of aspect declaration.
package foo; import org.aspectj.lang.ProceedingJoinPoint; import org.aspectj.lang.annotation.Aspect; import org.aspectj.lang.annotation.Around; import org.aspectj.lang.annotation.Pointcut; import org.springframework.util.StopWatch; import org.springframework.core.annotation.Order; @Aspect public class ProfilingAspect { @Around("methodsToBeProfiled()") public Object profile(ProceedingJoinPoint pjp) throws Throwable { StopWatch sw = new StopWatch(getClass().getSimpleName()); try { sw.start(pjp.getSignature().getName()); return pjp.proceed(); } finally { sw.stop(); System.out.println(sw.prettyPrint()); } } @Pointcut("execution(public * foo..*.*(..))") public void methodsToBeProfiled(){} }
We will also need to create an META-INF/aop.xml
file, to inform the AspectJ weaver
that we want to weave our ProfilingAspect
into our classes. This file convention,
namely the presence of a file (or files) on the Java classpath called
META-INF/aop.xml
is standard AspectJ.
<!DOCTYPE aspectj PUBLIC "-//AspectJ//DTD//EN" "http://www.eclipse.org/aspectj/dtd/aspectj.dtd"> <aspectj> <weaver> <!-- only weave classes in our application-specific packages --> <include within="foo.*"/> </weaver> <aspects> <!-- weave in just this aspect --> <aspect name="foo.ProfilingAspect"/> </aspects> </aspectj>
Now to the Spring-specific portion of the configuration. We need to configure a
LoadTimeWeaver
(all explained later, just take it on trust for now). This load-time
weaver is the essential component responsible for weaving the aspect configuration in
one or more META-INF/aop.xml
files into the classes in your application. The good
thing is that it does not require a lot of configuration, as can be seen below (there
are some more options that you can specify, but these are detailed later).
<?xml version="1.0" encoding="UTF-8"?> <beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:context="http://www.springframework.org/schema/context" xsi:schemaLocation=" http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context.xsd"> <!-- a service object; we will be profiling its methods --> <bean id="entitlementCalculationService" class="foo.StubEntitlementCalculationService"/> <!-- this switches on the load-time weaving --> <context:load-time-weaver/> </beans>
Now that all the required artifacts are in place - the aspect, the META-INF/aop.xml
file, and the Spring configuration -, let us create a simple driver class with a
main(..)
method to demonstrate the LTW in action.
package foo; import org.springframework.context.support.ClassPathXmlApplicationContext; public final class Main { public static void main(String[] args) { ApplicationContext ctx = new ClassPathXmlApplicationContext("beans.xml", Main.class); EntitlementCalculationService entitlementCalculationService = (EntitlementCalculationService) ctx.getBean("entitlementCalculationService"); // the profiling aspect is woven around this method execution entitlementCalculationService.calculateEntitlement(); } }
There is one last thing to do. The introduction to this section did say that one could
switch on LTW selectively on a per- ClassLoader
basis with Spring, and this is true.
However, just for this example, we are going to use a Java agent (supplied with Spring)
to switch on the LTW. This is the command line we will use to run the above Main
class:
java -javaagent:C:/projects/foo/lib/global/spring-instrument.jar foo.Main
The -javaagent
is a flag for specifying and enabling
agents
to instrument programs running on the JVM. The Spring Framework ships with such an
agent, the InstrumentationSavingAgent
, which is packaged in the
spring-instrument.jar
that was supplied as the value of the -javaagent
argument in
the above example.
The output from the execution of the Main
program will look something like that below.
(I have introduced a Thread.sleep(..)
statement into the calculateEntitlement()
implementation so that the profiler actually captures something other than 0
milliseconds - the 01234
milliseconds is not an overhead introduced by the AOP :) )
Calculating entitlement
StopWatch ProfilingAspect: running time (millis) = 1234
------ ----- ----------------------------
ms % Task name
------ ----- ----------------------------
01234 100% calculateEntitlement
Since this LTW is effected using full-blown AspectJ, we are not just limited to advising
Spring beans; the following slight variation on the Main
program will yield the same
result.
package foo; import org.springframework.context.support.ClassPathXmlApplicationContext; public final class Main { public static void main(String[] args) { new ClassPathXmlApplicationContext("beans.xml", Main.class); EntitlementCalculationService entitlementCalculationService = new StubEntitlementCalculationService(); // the profiling aspect will be woven around this method execution entitlementCalculationService.calculateEntitlement(); } }
Notice how in the above program we are simply bootstrapping the Spring container, and
then creating a new instance of the StubEntitlementCalculationService
totally outside
the context of Spring… the profiling advice still gets woven in.
The example admittedly is simplistic… however the basics of the LTW support in Spring have all been introduced in the above example, and the rest of this section will explain the why behind each bit of configuration and usage in detail.
Note | |
---|---|
The |
The aspects that you use in LTW have to be AspectJ aspects. They can be written in either the AspectJ language itself or you can write your aspects in the @AspectJ-style. It means that your aspects are then both valid AspectJ and Spring AOP aspects. Furthermore, the compiled aspect classes need to be available on the classpath.
The AspectJ LTW infrastructure is configured using one or more META-INF/aop.xml
files, that are on the Java classpath (either directly, or more typically in jar files).
The structure and contents of this file is detailed in the main AspectJ reference
documentation, and the interested reader is
referred to
that resource. (I appreciate that this section is brief, but the aop.xml
file is
100% AspectJ - there is no Spring-specific information or semantics that apply to it,
and so there is no extra value that I can contribute either as a result), so rather than
rehash the quite satisfactory section that the AspectJ developers wrote, I am just
directing you there.)
At a minimum you will need the following libraries to use the Spring Framework’s support for AspectJ LTW:
spring-aop.jar
(version 2.5 or later, plus all mandatory dependencies)
aspectjweaver.jar
(version 1.6.8 or later)
If you are using the Spring-provided agent to enable instrumentation, you will also need:
spring-instrument.jar
The key component in Spring’s LTW support is the LoadTimeWeaver
interface (in the
org.springframework.instrument.classloading
package), and the numerous implementations
of it that ship with the Spring distribution. A LoadTimeWeaver
is responsible for
adding one or more java.lang.instrument.ClassFileTransformers
to a ClassLoader
at
runtime, which opens the door to all manner of interesting applications, one of which
happens to be the LTW of aspects.
Tip | |
---|---|
If you are unfamiliar with the idea of runtime class file transformation, you are
encouraged to read the javadoc API documentation for the |
Configuring a LoadTimeWeaver
for a particular ApplicationContext
can be as easy as
adding one line. (Please note that you almost certainly will need to be using an
ApplicationContext
as your Spring container - typically a BeanFactory
will not be
enough because the LTW support makes use of BeanFactoryPostProcessors
.)
To enable the Spring Framework’s LTW support, you need to configure a LoadTimeWeaver
,
which typically is done using the @EnableLoadTimeWeaving
annotation.
@Configuration @EnableLoadTimeWeaving public class AppConfig { }
Alternatively, if you prefer XML based configuration, use the
<context:load-time-weaver/>
element. Note that the element is defined in the
context
namespace.
<?xml version="1.0" encoding="UTF-8"?> <beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:context="http://www.springframework.org/schema/context" xsi:schemaLocation=" http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context.xsd"> <context:load-time-weaver/> </beans>
The above configuration will define and register a number of LTW-specific infrastructure
beans for you automatically, such as a LoadTimeWeaver
and an AspectJWeavingEnabler
.
The default LoadTimeWeaver
is the DefaultContextLoadTimeWeaver
class, which attempts
to decorate an automatically detected LoadTimeWeaver
: the exact type of
LoadTimeWeaver
that will be automatically detected is dependent upon your runtime
environment (summarized in the following table).
Table 9.1. DefaultContextLoadTimeWeaver LoadTimeWeavers
Runtime Environment | LoadTimeWeaver implementation |
---|---|
Running in BEA’s Weblogic 10 |
|
Running in IBM WebSphere Application Server 7 |
|
Running in GlassFish |
|
Running in JBoss AS |
|
JVM started with Spring |
|
Fallback, expecting the underlying ClassLoader to follow common conventions (e.g.
applicable to |
|
Note that these are just the LoadTimeWeavers
that are autodetected when using the
DefaultContextLoadTimeWeaver
: it is of course possible to specify exactly which
LoadTimeWeaver
implementation that you wish to use.
To specify a specific LoadTimeWeaver
with Java configuration implement the
LoadTimeWeavingConfigurer
interface and override the getLoadTimeWeaver()
method:
@Configuration @EnableLoadTimeWeaving public class AppConfig implements LoadTimeWeavingConfigurer { @Override public LoadTimeWeaver getLoadTimeWeaver() { return new ReflectiveLoadTimeWeaver(); } }
If you are using XML based configuration you can specify the fully-qualified classname
as the value of the weaver-class
attribute on the <context:load-time-weaver/>
element:
<?xml version="1.0" encoding="UTF-8"?> <beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:context="http://www.springframework.org/schema/context" xsi:schemaLocation=" http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context.xsd"> <context:load-time-weaver weaver-class="org.springframework.instrument.classloading.ReflectiveLoadTimeWeaver"/> </beans>
The LoadTimeWeaver
that is defined and registered by the configuration can be later
retrieved from the Spring container using the well-known name loadTimeWeaver
.
Remember that the LoadTimeWeaver
exists just as a mechanism for Spring’s LTW
infrastructure to add one or more ClassFileTransformers
. The actual
ClassFileTransformer
that does the LTW is the ClassPreProcessorAgentAdapter
(from
the org.aspectj.weaver.loadtime
package) class. See the class-level javadocs of the
ClassPreProcessorAgentAdapter
class for further details, because the specifics of how
the weaving is actually effected is beyond the scope of this section.
There is one final attribute of the configuration left to discuss: the
aspectjWeaving
attribute (or aspectj-weaving
if you are using XML). This is a
simple attribute that controls whether LTW is enabled or not; it is as simple as that.
It accepts one of three possible values, summarized below, with the default value being
autodetect
if the attribute is not present.
Table 9.2. AspectJ weaving attribute values
Annotation Value | XML Value | Explanation |
---|---|---|
|
| AspectJ weaving is on, and aspects will be woven at load-time as appropriate. |
|
| LTW is off… no aspect will be woven at load-time. |
|
| If the Spring LTW infrastructure can find at least one |
This last section contains any additional settings and configuration that you will need when using Spring’s LTW support in environments such as application servers and web containers.
Apache Tomcat's default class loader does not support class
transformation which is why Spring provides an enhanced implementation that addresses
this need. Named TomcatInstrumentableClassLoader
, the loader works on Tomcat 5.0 and
above and can be registered individually for each web application as follows:
org.springframework.instrument.tomcat.jar
into $CATALINA_HOME/lib, where
$CATALINA_HOME represents the root of the Tomcat installation)
<Context path="/myWebApp" docBase="/my/webApp/location"> <Loader loaderClass="org.springframework.instrument.classloading.tomcat.TomcatInstrumentableClassLoader"/> </Context>
Apache Tomcat (6.0+) supports several context locations:
For efficiency, the embedded per-web-app configuration style is recommended because it will impact only applications that use the custom class loader and does not require any changes to the server configuration. See the Tomcat 6.0.x documentation for more details about available context locations.
Alternatively, consider the use of the Spring-provided generic VM agent, to be specified in Tomcat’s launch script (see above). This will make instrumentation available to all deployed web applications, no matter what ClassLoader they happen to run on.
Recent versions of WebLogic Server (version 10 and above), IBM WebSphere Application
Server (version 7 and above), Resin (3.1 and above) and JBoss (6.x or above) provide a
ClassLoader that is capable of local instrumentation. Spring’s native LTW leverages such
ClassLoaders to enable AspectJ weaving. You can enable LTW by simply activating
load-time weaving as described earlier. Specifically, you do not need to modify the
launch script to add -javaagent:path/to/spring-instrument.jar
.
Note that GlassFish instrumentation-capable ClassLoader is available only in its EAR environment. For GlassFish web applications, follow the Tomcat setup instructions as outlined above.
Note that on JBoss 6.x, the app server scanning needs to be disabled to prevent it from
loading the classes before the application actually starts. A quick workaround is to add
to your artifact a file named WEB-INF/jboss-scanning.xml
with the following content:
<scanning xmlns="urn:jboss:scanning:1.0"/>
When class instrumentation is required in environments that do not support or are not
supported by the existing LoadTimeWeaver
implementations, a JDK agent can be the only
solution. For such cases, Spring provides InstrumentationLoadTimeWeaver
, which
requires a Spring-specific (but very general) VM agent,
org.springframework.instrument-{version}.jar
(previously named spring-agent.jar
).
To use it, you must start the virtual machine with the Spring agent, by supplying the following JVM options:
-javaagent:/path/to/org.springframework.instrument-{version}.jar
Note that this requires modification of the VM launch script which may prevent you from using this in application server environments (depending on your operation policies). Additionally, the JDK agent will instrument the entire VM which can prove expensive.
For performance reasons, it is recommended to use this configuration only if your target environment (such as Jetty) does not have (or does not support) a dedicated LTW.
More information on AspectJ can be found on the AspectJ website.
The book Eclipse AspectJ by Adrian Colyer et. al. (Addison-Wesley, 2005) provides a comprehensive introduction and reference for the AspectJ language.
The book AspectJ in Action by Ramnivas Laddad (Manning, 2003) comes highly recommended; the focus of the book is on AspectJ, but a lot of general AOP themes are explored (in some depth).