Google Git
Sign in
eclipse / gerrit / www.eclipse.org / aspectj / 9cc5e17c86c9f00b82b99e1ad6d40acae0af50f4 / . / doc / next / progguide / printable.html
blob: 820221c7d3a81112a9fca8727d5cfaf11c80d594 [file] [log] [blame]
<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<title>The AspectJTM Programming Guide</title><link rel="stylesheet" href="aspectj-docs.css" type="text/css"><meta name="generator" content="DocBook XSL Stylesheets V1.44"></head><body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF"><div class="book" id="d0e1"><div class="titlepage"><div><h1 class="title"><a name="d0e1"></a>The AspectJ<sup>TM</sup> Programming Guide</h1></div><div><h3 class="author">the AspectJ Team</h3></div><div><div class="legalnotice"><p>
Copyright (c) 1998-2001 Xerox Corporation,
2002-2003 Palo Alto Research Center, Incorporated.
All rights reserved.
</p></div></div><div><div class="abstract"><p><a name="d0e15"></a><b>Abstract</b></p><p>
This programming guide describes the AspectJ language. A
companion guide describes the tools which are part of the
AspectJ development environment.
</p><p>
If you are completely new to AspectJ, you should first read
<a href="#starting">Getting Started with AspectJ</a> for a broad overview of programming
in AspectJ. If you are already familiar with AspectJ, but want a deeper
understanding, you should read <a href="#language">The AspectJ Language</a> and
look at the examples in the chapter. If you want a more formal
definition of AspectJ, you should read <a href="#semantics">Semantics</a>.
</p></div></div><hr></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#preface">Preface</a></dt><dt>1. <a href="#starting">Getting Started with AspectJ</a></dt><dd><dl><dt><a href="#starting-intro">Introduction</a></dt><dt><a href="#starting-aspectj">Introduction to AspectJ</a></dt><dd><dl><dt><a href="#the-dynamic-join-point-model">The Dynamic Join Point Model</a></dt><dt><a href="#pointcuts">Pointcuts</a></dt><dt><a href="#advice">Advice</a></dt><dt><a href="#inter-type-declarations">Inter-type declarations</a></dt><dt><a href="#aspects">Aspects</a></dt></dl></dd><dt><a href="#starting-development">Development Aspects</a></dt><dd><dl><dt><a href="#tracing">Tracing</a></dt><dt><a href="#profiling-and-logging">Profiling and Logging</a></dt><dt><a href="#pre-and-post-conditions">Pre- and Post-Conditions</a></dt><dt><a href="#contract-enforcement">Contract Enforcement</a></dt><dt><a href="#configuration-management">Configuration Management</a></dt></dl></dd><dt><a href="#starting-production">Production Aspects</a></dt><dd><dl><dt><a href="#change-monitoring">Change Monitoring</a></dt><dt><a href="#context-passing">Context Passing</a></dt><dt><a href="#starting-production-consistentBehavior">Providing Consistent Behavior</a></dt></dl></dd><dt><a href="#starting-conclusion">Conclusion</a></dt></dl></dd><dt>2. <a href="#language">The AspectJ Language</a></dt><dd><dl><dt><a href="#language-intro">Introduction</a></dt><dt><a href="#language-anatomy">The Anatomy of an Aspect</a></dt><dd><dl><dt><a href="#an-example-aspect">An Example Aspect</a></dt><dt><a href="#pointcuts">Pointcuts</a></dt><dt><a href="#advice">Advice</a></dt></dl></dd><dt><a href="#language-joinPoints">Join Points and Pointcuts</a></dt><dd><dl><dt><a href="#some-example-pointcuts">Some Example Pointcuts</a></dt><dt><a href="#call-vs-execution">call vs. execution</a></dt><dt><a href="#pointcut-composition">Pointcut composition</a></dt><dt><a href="#pointcut-parameters">Pointcut Parameters</a></dt><dt><a href="#example">Example: <tt>HandleLiveness</tt></a></dt></dl></dd><dt><a href="#language-advice">Advice</a></dt><dt><a href="#language-interType">Inter-type declarations</a></dt><dd><dl><dt><a href="#inter-type-scope">Inter-type Scope</a></dt><dt><a href="#example-pointassertions">Example: <tt>PointAssertions</tt></a></dt></dl></dd><dt><a href="#language-thisJoinPoint">thisJoinPoint</a></dt></dl></dd><dt>3. <a href="#examples">Examples</a></dt><dd><dl><dt><a href="#examples-intro">Introduction</a></dt><dt><a href="#examples-howto">Obtaining, Compiling and Running the Examples</a></dt><dt><a href="#examples-basic">Basic Techniques</a></dt><dd><dl><dt><a href="#examples-joinPoints">Join Points and <tt>thisJoinPoint</tt></a></dt><dt><a href="#examples-roles">Roles and Views</a></dt></dl></dd><dt><a href="#examples-development">Development Aspects</a></dt><dd><dl><dt><a href="#tracing-using-aspects">Tracing using aspects</a></dt></dl></dd><dt><a href="#examples-production">Production Aspects</a></dt><dd><dl><dt><a href="#a-bean-aspect">A Bean Aspect</a></dt><dt><a href="#the-subject-observer-protocol">The Subject/Observer Protocol</a></dt><dt><a href="#a-simple-telecom-simulation">A Simple Telecom Simulation</a></dt></dl></dd><dt><a href="#examples-reusable">Reusable Aspects</a></dt><dd><dl><dt><a href="#tracing-using-aspects-revisited">Tracing using Aspects, Revisited</a></dt></dl></dd></dl></dd><dt>4. <a href="#idioms">Idioms</a></dt><dd><dl><dt><a href="#idioms-intro">Introduction</a></dt></dl></dd><dt>5. <a href="#pitfalls">Pitfalls</a></dt><dd><dl><dt><a href="#pitfalls-intro">Introduction</a></dt><dt><a href="#pitfalls-infiniteLoops">Infinite loops</a></dt></dl></dd><dt>A. <a href="#quick">AspectJ Quick Reference</a></dt><dd><dl><dt><a href="#quick-pointcuts">Pointcuts</a></dt><dt><a href="#quick-typePatterns">Type Patterns</a></dt><dt><a href="#quick-advice">Advice</a></dt><dt><a href="#quick-interType">Inter-type member declarations</a></dt><dt><a href="#quick-other">Other declarations</a></dt><dt><a href="#quick-aspectAssociations">Aspects</a></dt></dl></dd><dt>B. <a href="#semantics">Language Semantics</a></dt><dd><dl><dt><a href="#semantics-intro">Introduction</a></dt><dt><a href="#semantics-joinPoints">Join Points</a></dt><dt><a href="#semantics-pointcuts">Pointcuts</a></dt><dd><dl><dt><a href="#pointcut-definition">Pointcut definition</a></dt><dt><a href="#context-exposure">Context exposure</a></dt><dt><a href="#primitive-pointcuts">Primitive pointcuts</a></dt><dt><a href="#signatures">Signatures</a></dt><dt><a href="#matching">Matching</a></dt><dt><a href="#type-patterns">Type patterns</a></dt><dt><a href="#pattern-summary">Pattern Summary</a></dt></dl></dd><dt><a href="#semantics-advice">Advice</a></dt><dd><dl><dt><a href="#advice-modifiers">Advice modifiers</a></dt><dt><a href="#advice-and-checked-exceptions">Advice and checked exceptions</a></dt><dt><a href="#advice-precedence">Advice precedence</a></dt><dt><a href="#reflective-access-to-the-join-point">Reflective access to the join point</a></dt></dl></dd><dt><a href="#semantics-declare">Static crosscutting</a></dt><dd><dl><dt><a href="#inter-type-member-declarations">Inter-type member declarations</a></dt><dt><a href="#access-modifiers">Access modifiers</a></dt><dt><a href="#conflicts">Conflicts</a></dt><dt><a href="#extension-and-implementation">Extension and Implementation</a></dt><dt><a href="#interfaces-with-members">Interfaces with members</a></dt><dt><a href="#warnings-and-errors">Warnings and Errors</a></dt><dt><a href="#softened-exceptions">Softened exceptions</a></dt><dt><a href="#advice-precedence">Advice Precedence</a></dt><dt><a href="#statically-determinable-pointcuts">Statically determinable pointcuts</a></dt></dl></dd><dt><a href="#semantics-aspects">Aspects</a></dt><dd><dl><dt><a href="#aspect-declaration">Aspect Declaration</a></dt><dt><a href="#aspect-extension">Aspect Extension</a></dt><dt><a href="#aspect-instantiation">Aspect instantiation</a></dt><dt><a href="#aspect-privilege">Aspect privilege</a></dt></dl></dd></dl></dd><dt>C. <a href="#implementation">Implementation Notes</a></dt><dd><dl><dt><a href="#d0e7185">Compiler Notes</a></dt><dt><a href="#d0e7259">Bytecode Notes</a></dt><dd><dl><dt><a href="#the-class-expression-and-string">The .class expression and String +</a></dt><dt><a href="#the-handler-join-point">The Handler join point</a></dt><dt><a href="#initializers-and-inter-type-constructors">Initializers and Inter-type Constructors</a></dt></dl></dd><dt><a href="#d0e7365">Summary of implementation requirements</a></dt></dl></dd></dl></div><div id="preface" class="preface"><div class="titlepage"><div><h2 class="title"><a name="preface"></a>Preface</h2></div></div><p>
This programming guide does three things. It
<div class="itemizedlist"><ul compact><li><p><a name="d0e32"></a>introduces the AspectJ language</p></li><li><p><a name="d0e35"></a>
defines each of AspectJ's constructs and their semantics, and
</p></li><li><p><a name="d0e38"></a>
provides examples of their use.
</p></li></ul></div>
It includes appendices that give a reference to the syntax of AspectJ,
a more formal description of AspectJ's semantics, and a description of
notes about the AspectJ implementation.
</p><p>
The first section, <a href="#starting">Getting Started with AspectJ</a>, provides a gentle
overview of writing AspectJ programs. It also shows how one can
introduce AspectJ into an existing development effort in stages,
reducing the associated risk. You should read this section if this is
your first exposure to AspectJ and you want to get a sense of what
AspectJ is all about.
</p><p>
The second section, <a href="#language">The AspectJ Language</a>, covers the features of
the language in more detail, using code snippets as examples. All the
basics of the language is covered, and after reading this section, you
should be able to use the language correctly.
</p><p>
The next section, <a href="#examples">Examples</a>, comprises a set of
complete programs that not only show the features being used, but also
try to illustrate recommended practice. You should read this section
after you are familiar with the elements of AspectJ.
</p><p>
Finally, there are two short chapters, one on <a href="#idioms">Idioms</a>
and one on <a href="#pitfalls">Pitfalls</a>.
</p><p>
The back matter contains several appendices that cover a <a href="#quick">AspectJ Quick Reference</a> to the language's syntax, a more
in depth coverage of its <a href="#semantics">Semantics</a>,
and a description of the latitude enjoyed by its <a href="#implementation">Implementation Notes</a>.
</p></div><div class="chapter"><div class="titlepage"><div><h2 class="title"><a name="starting"></a>Chapter 1. Getting Started with AspectJ</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#starting-intro">Introduction</a></dt><dt><a href="#starting-aspectj">Introduction to AspectJ</a></dt><dd><dl><dt><a href="#the-dynamic-join-point-model">The Dynamic Join Point Model</a></dt><dt><a href="#pointcuts">Pointcuts</a></dt><dt><a href="#advice">Advice</a></dt><dt><a href="#inter-type-declarations">Inter-type declarations</a></dt><dt><a href="#aspects">Aspects</a></dt></dl></dd><dt><a href="#starting-development">Development Aspects</a></dt><dd><dl><dt><a href="#tracing">Tracing</a></dt><dt><a href="#profiling-and-logging">Profiling and Logging</a></dt><dt><a href="#pre-and-post-conditions">Pre- and Post-Conditions</a></dt><dt><a href="#contract-enforcement">Contract Enforcement</a></dt><dt><a href="#configuration-management">Configuration Management</a></dt></dl></dd><dt><a href="#starting-production">Production Aspects</a></dt><dd><dl><dt><a href="#change-monitoring">Change Monitoring</a></dt><dt><a href="#context-passing">Context Passing</a></dt><dt><a href="#starting-production-consistentBehavior">Providing Consistent Behavior</a></dt></dl></dd><dt><a href="#starting-conclusion">Conclusion</a></dt></dl></div><div class="sect1"><a name="starting-intro"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="starting-intro"></a>Introduction</h2></div></div><p>
Many software developers are attracted to the idea of aspect-oriented
programming (AOP) but unsure about how to begin using the
technology. They recognize the concept of crosscutting concerns, and
know that they have had problems with the implementation of such
concerns in the past. But there are many questions about how to adopt
AOP into the development process. Common questions include:
<div class="itemizedlist"><ul compact><li><p><a name="d0e80"></a>Can I use aspects in my existing code?</p></li><li><p><a name="d0e83"></a>
What kinds of benefits can I expect to get from using aspects?
</p></li><li><p><a name="d0e86"></a>How do I find aspects in my programs?</p></li><li><p><a name="d0e89"></a>How steep is the learning curve for AOP?</p></li><li><p><a name="d0e92"></a>What are the risks of using this new technology?</p></li></ul></div>
</p><p>
This chapter addresses these questions in the context of AspectJ: a
general-purpose aspect-oriented extension to Java. A series of
abridged examples illustrate the kinds of aspects programmers may
want to implement using AspectJ and the benefits associated with
doing so. Readers who would like to understand the examples in more
detail, or who want to learn how to program examples like these, can
find more complete examples and supporting material linked from the
AspectJ web site ( <a href="http://eclipse.org/aspectj" target="_top">http://eclipse.org/aspectj</a> ).
</p><p>
A significant risk in adopting any new technology is going too far
too fast. Concern about this risk causes many organizations to be
conservative about adopting new technology. To address this issue,
the examples in this chapter are grouped into three broad categories,
with aspects that are easier to adopt into existing development
projects coming earlier in this chapter. The next section, <a href="#starting-aspectj">Introduction to AspectJ</a>, we present the core of AspectJ's
features, and in <a href="#starting-development">Development Aspects</a>, we present
aspects that facilitate tasks such as debugging, testing and
performance tuning of applications. And, in the section following,
<a href="#starting-production">Production Aspects</a>, we present aspects that
implement crosscutting functionality common in Java applications. We
will defer discussing a third category of aspects, reusable aspects,
until <a href="#language">The AspectJ Language</a>.
</p><p>
These categories are informal, and this ordering is not the only way
to adopt AspectJ. Some developers may want to use a production aspect
right away. But our experience with current AspectJ users suggests
that this is one ordering that allows developers to get experience
with (and benefit from) AOP technology quickly, while also minimizing
risk.
</p></div><div class="sect1"><a name="starting-aspectj"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="starting-aspectj"></a>Introduction to AspectJ</h2></div></div><p>
This section presents a brief introduction to the features of AspectJ
used later in this chapter. These features are at the core of the
language, but this is by no means a complete overview of AspectJ.
</p><p>
The features are presented using a simple figure editor system. A
<tt>Figure</tt> consists of a number of
<tt>FigureElements</tt>, which can be either
<tt>Point</tt>s or <tt>Line</tt>s. The
<tt>Figure</tt> class provides factory services. There
is also a <tt>Display</tt>. Most example programs later
in this chapter are based on this system as well.
</p><p>
<div class="mediaobject"><img src="figureUML.gif"><div class="caption"><p>
UML for the <tt>FigureEditor</tt> example
</p></div></div>
</p><p>
The motivation for AspectJ (and likewise for aspect-oriented
programming) is the realization that there are issues or concerns
that are not well captured by traditional programming
methodologies. Consider the problem of enforcing a security policy in
some application. By its nature, security cuts across many of the
natural units of modularity of the application. Moreover, the
security policy must be uniformly applied to any additions as the
application evolves. And the security policy that is being applied
might itself evolve. Capturing concerns like a security policy in a
disciplined way is difficult and error-prone in a traditional
programming language.
</p><p>
Concerns like security cut across the natural units of
modularity. For object-oriented programming languages, the natural
unit of modularity is the class. But in object-oriented programming
languages, crosscutting concerns are not easily turned into classes
precisely because they cut across classes, and so these aren't
reusable, they can't be refined or inherited, they are spread through
out the program in an undisciplined way, in short, they are difficult
to work with.
</p><p>
Aspect-oriented programming is a way of modularizing crosscutting
concerns much like object-oriented programming is a way of
modularizing common concerns. AspectJ is an implementation of
aspect-oriented programming for Java.
</p><p>
AspectJ adds to Java just one new concept, a join point -- and that's
really just a name for an existing Java concept. It adds to Java
only a few new constructs: pointcuts, advice, inter-type declarations
and aspects. Pointcuts and advice dynamically affect program flow,
inter-type declarations statically affects a program's class
hierarchy, and aspects encapsulate these new constructs.
</p><p>
A <span class="emphasis"><i>join point</i></span> is a well-defined point in the
program flow. A <span class="emphasis"><i>pointcut</i></span> picks out certain join
points and values at those points. A piece of
<span class="emphasis"><i>advice</i></span> is code that is executed when a join
point is reached. These are the dynamic parts of AspectJ.
</p><p>
AspectJ also has different kinds of <span class="emphasis"><i>inter-type
declarations</i></span> that allow the programmer to modify a
program's static structure, namely, the members of its classes and
the relationship between classes.
</p><p>
AspectJ's <span class="emphasis"><i>aspect</i></span> are the unit of modularity for
crosscutting concerns. They behave somewhat like Java classes, but
may also include pointcuts, advice and inter-type declarations.
</p><p>
In the sections immediately following, we are first going to look at
join points and how they compose into pointcuts. Then we will look at
advice, the code which is run when a pointcut is reached. We will see
how to combine pointcuts and advice into aspects, AspectJ's reusable,
inheritable unit of modularity. Lastly, we will look at how to use
inter-type declarations to deal with crosscutting concerns of a
program's class structure.
</p><div class="sect2"><a name="the-dynamic-join-point-model"></a><div class="titlepage"><div><h3 class="title"><a name="the-dynamic-join-point-model"></a>The Dynamic Join Point Model</h3></div></div><p>
A critical element in the design of any aspect-oriented language is
the join point model. The join point model provides the common
frame of reference that makes it possible to define the dynamic
structure of crosscutting concerns. This chapter describes
AspectJ's dynamic join points, in which join points are certain
well-defined points in the execution of the program.
</p><p>
AspectJ provides for many kinds of join points, but this chapter
discusses only one of them: method call join points. A method call
join point encompasses the actions of an object receiving a method
call. It includes all the actions that comprise a method call,
starting after all arguments are evaluated up to and including
return (either normally or by throwing an exception).
</p><p>
Each method call at runtime is a different join point, even if it
comes from the same call expression in the program. Many other
join points may run while a method call join point is executing --
all the join points that happen while executing the method body,
and in those methods called from the body. We say that these join
points execute in the <span class="emphasis"><i>dynamic context</i></span> of the
original call join point.
</p></div><div class="sect2"><a name="pointcuts"></a><div class="titlepage"><div><h3 class="title"><a name="pointcuts"></a>Pointcuts</h3></div></div><p>
In AspectJ, <span class="emphasis"><i>pointcuts</i></span> pick out certain join
points in the program flow. For example, the pointcut
</p><pre class="programlisting">
call(void Point.setX(int))
</pre><p>
picks out each join point that is a call to a method that has the
signature <tt>void Point.setX(int)</tt> &#8212; that is,
<tt>Point</tt>'s void <tt>setX</tt>
method with a single <tt>int</tt> parameter.
</p><p>
A pointcut can be built out of other pointcuts with and, or, and
not (spelled <tt>&amp;&amp;</tt>, <tt>||</tt>,
and <tt>!</tt>). For example:
</p><pre class="programlisting">
call(void Point.setX(int)) ||
call(void Point.setY(int))
</pre><p>
picks out each join point that is either a call to
<tt>setX</tt> or a call to <tt>setY</tt>.
</p><p>
Pointcuts can identify join points from many different types
&#8212; in other words, they can crosscut types. For example,
</p><pre class="programlisting">
call(void FigureElement.setXY(int,int)) ||
call(void Point.setX(int)) ||
call(void Point.setY(int)) ||
call(void Line.setP1(Point)) ||
call(void Line.setP2(Point));
</pre><p>
picks out each join point that is a call to one of five methods
(the first of which is an interface method, by the way).
</p><p>
In our example system, this pointcut captures all the join points
when a <tt>FigureElement</tt> moves. While this is a
useful way to specify this crosscutting concern, it is a bit of a
mouthful. So AspectJ allows programmers to define their own named
pointcuts with the <tt>pointcut</tt> form. So the
following declares a new, named pointcut:
</p><pre class="programlisting">
pointcut move():
call(void FigureElement.setXY(int,int)) ||
call(void Point.setX(int)) ||
call(void Point.setY(int)) ||
call(void Line.setP1(Point)) ||
call(void Line.setP2(Point));
</pre><p>
and whenever this definition is visible, the programmer can simply
use <tt>move()</tt> to capture this complicated
pointcut.
</p><p>
The previous pointcuts are all based on explicit enumeration of a
set of method signatures. We sometimes call this
<span class="emphasis"><i>name-based</i></span> crosscutting. AspectJ also
provides mechanisms that enable specifying a pointcut in terms of
properties of methods other than their exact name. We call this
<span class="emphasis"><i>property-based</i></span> crosscutting. The simplest of
these involve using wildcards in certain fields of the method
signature. For example, the pointcut
</p><pre class="programlisting">
call(void Figure.make*(..))
</pre><p>
picks out each join point that's a call to a void method defined
on <tt>Figure</tt> whose the name begins with
"<tt>make</tt>" regardless of the method's parameters.
In our system, this picks out calls to the factory methods
<tt>makePoint</tt> and <tt>makeLine</tt>.
The pointcut
</p><pre class="programlisting">
call(public * Figure.* (..))
</pre><p>
picks out each call to <tt>Figure</tt>'s public
methods.
</p><p>
But wildcards aren't the only properties AspectJ supports.
Another pointcut, <tt>cflow</tt>, identifies join
points based on whether they occur in the dynamic context of
other join points. So
</p><pre class="programlisting">
cflow(move())
</pre><p>
picks out each join point that occurs in the dynamic context of
the join points picked out by <tt>move()</tt>, our named
pointcut defined above. So this picks out each join points that
occurrs between when a move method is called and when it returns
(either normally or by throwing an exception).
</p></div><div class="sect2"><a name="advice"></a><div class="titlepage"><div><h3 class="title"><a name="advice"></a>Advice</h3></div></div><p>
So pointcuts pick out join points. But they don't
<span class="emphasis"><i>do</i></span> anything apart from picking out join
points. To actually implement crosscutting behavior, we use
advice. Advice brings together a pointcut (to pick out join
points) and a body of code (to run at each of those join points).
</p><p>
AspectJ has several different kinds of advice. <span class="emphasis"><i>Before
advice</i></span> runs as a join point is reached, before the
program proceeds with the join point. For example, before advice
on a method call join point runs before the actual method starts
running, just after the arguments to the method call are evaluated.
</p><pre class="programlisting">
before(): move() {
System.out.println("about to move");
}
</pre><p>
<span class="emphasis"><i>After advice</i></span> on a particular join point runs
after the program proceeds with that join point. For example,
after advice on a method call join point runs after the method body
has run, just before before control is returned to the caller.
Because Java programs can leave a join point 'normally' or by
throwing an exception, there are three kinds of after advice:
<tt>after returning</tt>, <tt>after
throwing</tt>, and plain <tt>after</tt> (which runs
after returning <span class="emphasis"><i>or</i></span> throwing, like Java's
<tt>finally</tt>).
</p><pre class="programlisting">
after() returning: move() {
System.out.println("just successfully moved");
}
</pre><p>
<span class="emphasis"><i>Around advice</i></span> on a join point runs as the join
point is reached, and has explicit control over whether the program
proceeds with the join point. Around advice is not discussed in
this section.
</p><div class="sect3"><a name="d0e346"></a><div class="titlepage"><div><h4 class="title"><a name="d0e346"></a>Exposing Context in Pointcuts</h4></div></div><p>
Pointcuts not only pick out join points, they can also expose
part of the execution context at their join points. Values
exposed by a pointcut can be used in the body of advice
declarations.
</p><p>
An advice declaration has a parameter list (like a method) that
gives names to all the pieces of context that it uses. For
example, the after advice
</p><pre class="programlisting">
after(FigureElement fe, int x, int y) returning:
...SomePointcut... {
...SomeBody...
}
</pre><p>
uses three pieces of exposed context, a
<tt>FigureElement</tt> named fe, and two
<tt>int</tt>s named x and y.
</p><p>
The body of the advice uses the names just like method
parameters, so
</p><pre class="programlisting">
after(FigureElement fe, int x, int y) returning:
...SomePointcut... {
System.out.println(fe + " moved to (" + x + ", " + y + ")");
}
</pre><p>
The advice's pointcut publishes the values for the advice's
arguments. The three primitive pointcuts
<tt>this</tt>, <tt>target</tt> and
<tt>args</tt> are used to publish these values. So now
we can write the complete piece of advice:
</p><pre class="programlisting">
after(FigureElement fe, int x, int y) returning:
call(void FigureElement.setXY(int, int))
&amp;&amp; target(fe)
&amp;&amp; args(x, y) {
System.out.println(fe + " moved to (" + x + ", " + y + ")");
}
</pre><p>
The pointcut exposes three values from calls to
<tt>setXY</tt>: the target
<tt>FigureElement</tt> -- which it publishes as
<tt>fe</tt>, so it becomes the first argument to the
after advice -- and the two int arguments -- which it publishes
as <tt>x</tt> and <tt>y</tt>, so they become
the second and third argument to the after advice.
</p><p>
So the advice prints the figure element
that was moved and its new <tt>x</tt> and
<tt>y</tt> coordinates after each
<tt>setXY</tt> method call.
</p><p>
A named pointcut may have parameters like a piece of advice.
When the named pointcut is used (by advice, or in another named
pointcut), it publishes its context by name just like the
<tt>this</tt>, <tt>target</tt> and
<tt>args</tt> pointcut. So another way to write the
above advice is
</p><pre class="programlisting">
pointcut setXY(FigureElement fe, int x, int y):
call(void FigureElement.setXY(int, int))
&amp;&amp; target(fe)
&amp;&amp; args(x, y);
after(FigureElement fe, int x, int y) returning: setXY(fe, x, y) {
System.out.println(fe + " moved to (" + x + ", " + y + ").");
}
</pre></div></div><div class="sect2"><a name="inter-type-declarations"></a><div class="titlepage"><div><h3 class="title"><a name="inter-type-declarations"></a>Inter-type declarations</h3></div></div><p>
Inter-type declarations in AspectJ are declarations that cut across
classes and their hierarchies. They may declare members that cut
across multiple classes, or change the inheritance relationship
between classes. Unlike advice, which operates primarily
dynamically, introduction operates statically, at compile-time.
</p><p>
Consider the problem of expressing a capability shared by some
existing classes that are already part of a class hierarchy,
i.e. they already extend a class. In Java, one creates an
interface that captures this new capability, and then adds to
<span class="emphasis"><i>each affected class</i></span> a method that implements
this interface.
</p><p>
AspectJ can express the concern in one place, by using inter-type
declarations. The aspect declares the methods and fields that are
necessary to implement the new capability, and associates the
methods and fields to the existing classes.
</p><p>
Suppose we want to have <tt>Screen</tt> objects
observe changes to <tt>Point</tt> objects, where
<tt>Point</tt> is an existing class. We can implement
this by writing an aspect declaring that the class Point
<tt>Point</tt> has an instance field,
<tt>observers</tt>, that keeps track of the
<tt>Screen</tt> objects that are observing
<tt>Point</tt>s.
</p><pre class="programlisting">
aspect PointObserving {
private Vector Point.observers = new Vector();
...
}
</pre><p>
The <tt>observers</tt> field is private, so only
<tt>PointObserving</tt> can see it. So observers are
added or removed with the static methods
<tt>addObserver</tt> and
<tt>removeObserver</tt> on the aspect.
</p><pre class="programlisting">
aspect PointObserving {
private Vector Point.observers = new Vector();
public static void addObserver(Point p, Screen s) {
p.observers.add(s);
}
public static void removeObserver(Point p, Screen s) {
p.observers.remove(s);
}
...
}
</pre><p>
Along with this, we can define a pointcut
<tt>changes</tt> that defines what we want to observe,
and the after advice defines what we want to do when we observe a
change.
</p><pre class="programlisting">
aspect PointObserving {
private Vector Point.observers = new Vector();
public static void addObserver(Point p, Screen s) {
p.observers.add(s);
}
public static void removeObserver(Point p, Screen s) {
p.observers.remove(s);
}
pointcut changes(Point p): target(p) &amp;&amp; call(void Point.set*(int));
after(Point p): changes(p) {
Iterator iter = p.observers.iterator();
while ( iter.hasNext() ) {
updateObserver(p, (Screen)iter.next());
}
}
static void updateObserver(Point p, Screen s) {
s.display(p);
}
}
</pre><p>
Note that neither <tt>Screen</tt>'s nor
<tt>Point</tt>'s code has to be modified, and that
all the changes needed to support this new capability are local to
this aspect.
</p></div><div class="sect2"><a name="aspects"></a><div class="titlepage"><div><h3 class="title"><a name="aspects"></a>Aspects</h3></div></div><p>
Aspects wrap up pointcuts, advice, and inter-type declarations in a
a modular unit of crosscutting implementation. It is defined very
much like a class, and can have methods, fields, and initializers
in addition to the crosscutting members. Because only aspects may
include these crosscutting members, the declaration of these
effects is localized.
</p><p>
Like classes, aspects may be instantiated, but AspectJ controls how
that instantiation happens -- so you can't use Java's
<tt>new</tt> form to build new aspect instances. By
default, each aspect is a singleton, so one aspect instance is
created. This means that advice may use non-static fields of the
aspect, if it needs to keep state around:
</p><pre class="programlisting">
aspect Logging {
OutputStream logStream = System.err;
before(): move() {
logStream.println("about to move");
}
}
</pre><p>
Aspects may also have more complicated rules for instantiation, but
these will be described in a later chapter.
</p></div></div><div class="sect1"><a name="starting-development"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="starting-development"></a>Development Aspects</h2></div></div><p>
The next two sections present the use of aspects in increasingly
sophisticated ways. Development aspects are easily removed from
production builds. Production aspects are intended to be used in
both development and in production, but tend to affect only a few
classes.
</p><p>
This section presents examples of aspects that can be used during
development of Java applications. These aspects facilitate debugging,
testing and performance tuning work. The aspects define behavior that
ranges from simple tracing, to profiling, to testing of internal
consistency within the application. Using AspectJ makes it possible
to cleanly modularize this kind of functionality, thereby making it
possible to easily enable and disable the functionality when desired.
</p><div class="sect2"><a name="tracing"></a><div class="titlepage"><div><h3 class="title"><a name="tracing"></a>Tracing</h3></div></div><p>
This first example shows how to increase the visibility of the
internal workings of a program. It is a simple tracing aspect that
prints a message at specified method calls. In our figure editor
example, one such aspect might simply trace whenever points are
drawn.
</p><pre class="programlisting">
aspect SimpleTracing {
pointcut tracedCall():
call(void FigureElement.draw(GraphicsContext));
before(): tracedCall() {
System.out.println("Entering: " + thisJoinPoint);
}
}
</pre><p>
This code makes use of the <tt>thisJoinPoint</tt> special
variable. Within all advice bodies this variable is bound to an
object that describes the current join point. The effect of this
code is to print a line like the following every time a figure
element receives a <tt>draw</tt> method call:
</p><pre class="programlisting">
Entering: call(void FigureElement.draw(GraphicsContext))
</pre><p>
To understand the benefit of coding this with AspectJ consider
changing the set of method calls that are traced. With AspectJ,
this just requires editing the definition of the
<tt>tracedCalls</tt> pointcut and recompiling. The
individual methods that are traced do not need to be edited.
</p><p>
When debugging, programmers often invest considerable effort in
figuring out a good set of trace points to use when looking for a
particular kind of problem. When debugging is complete or appears
to be complete it is frustrating to have to lose that investment by
deleting trace statements from the code. The alternative of just
commenting them out makes the code look bad, and can cause trace
statements for one kind of debugging to get confused with trace
statements for another kind of debugging.
</p><p>
With AspectJ it is easy to both preserve the work of designing a
good set of trace points and disable the tracing when it isn t
being used. This is done by writing an aspect specifically for that
tracing mode, and removing that aspect from the compilation when it
is not needed.
</p><p>
This ability to concisely implement and reuse debugging
configurations that have proven useful in the past is a direct
result of AspectJ modularizing a crosscutting design element the
set of methods that are appropriate to trace when looking for a
given kind of information.
</p></div><div class="sect2"><a name="profiling-and-logging"></a><div class="titlepage"><div><h3 class="title"><a name="profiling-and-logging"></a>Profiling and Logging</h3></div></div><p>
Our second example shows you how to do some very specific
profiling. Although many sophisticated profiling tools are
available, and these can gather a variety of information and
display the results in useful ways, you may sometimes want to
profile or log some very specific behavior. In these cases, it is
often possible to write a simple aspect similar to the ones above
to do the job.
</p><p>
For example, the following aspect counts the number of calls to the
<tt>rotate</tt> method on a <tt>Line</tt>
and the number of calls to the <tt>set*</tt> methods of
a <tt>Point</tt> that happen within the control flow
of those calls to <tt>rotate</tt>:
</p><pre class="programlisting">
aspect SetsInRotateCounting {
int rotateCount = 0;
int setCount = 0;
before(): call(void Line.rotate(double)) {
rotateCount++;
}
before(): call(void Point.set*(int))
&amp;&amp; cflow(call(void Line.rotate(double))) {
setCount++;
}
}
</pre><p>
In effect, this aspect allows the programmer to ask very specific
questions like
<blockquote class="blockquote">
How many times is the <tt>rotate</tt>
method defined on <tt>Line</tt> objects called?
</blockquote>
and
<blockquote class="blockquote">
How many times are methods defined on
<tt>Point</tt> objects whose name begins with
"<tt>set</tt>" called in fulfilling those rotate
calls?
</blockquote>
questions it may be difficult to express using standard
profiling or logging tools.
</p></div><div class="sect2"><a name="pre-and-post-conditions"></a><div class="titlepage"><div><h3 class="title"><a name="pre-and-post-conditions"></a>Pre- and Post-Conditions</h3></div></div><p>
Many programmers use the "Design by Contract" style popularized by
Bertand Meyer in <i>Object-Oriented Software Construction,
2/e</i>. In this style of programming, explicit
pre-conditions test that callers of a method call it properly and
explicit post-conditions test that methods properly do the work
they are supposed to.
</p><p>
AspectJ makes it possible to implement pre- and post-condition
testing in modular form. For example, this code
</p><pre class="programlisting">
aspect PointBoundsChecking {
pointcut setX(int x):
(call(void FigureElement.setXY(int, int)) &amp;&amp; args(x, *))
|| (call(void Point.setX(int)) &amp;&amp; args(x));
pointcut setY(int y):
(call(void FigureElement.setXY(int, int)) &amp;&amp; args(*, y))
|| (call(void Point.setY(int)) &amp;&amp; args(y));
before(int x): setX(x) {
if ( x &lt; MIN_X || x &gt; MAX_X )
throw new IllegalArgumentException("x is out of bounds.");
}
before(int y): setY(y) {
if ( y &lt; MIN_Y || y &gt; MAX_Y )
throw new IllegalArgumentException("y is out of bounds.");
}
}
</pre><p>
implements the bounds checking aspect of pre-condition testing for
operations that move points. Notice that the
<tt>setX</tt> pointcut refers to all the operations
that can set a Point's <tt>x</tt> coordinate; this
includes the <tt>setX</tt> method, as well as half of
the <tt>setXY</tt> method. In this sense the
<tt>setX</tt> pointcut can be seen as involving very
fine-grained crosscutting &#8212; it names the the
<tt>setX</tt> method and half of the
<tt>setXY</tt> method.
</p><p>
Even though pre- and post-condition testing aspects can often be
used only during testing, in some cases developers may wish to
include them in the production build as well. Again, because
AspectJ makes it possible to modularize these crosscutting concerns
cleanly, it gives developers good control over this decision.
</p></div><div class="sect2"><a name="contract-enforcement"></a><div class="titlepage"><div><h3 class="title"><a name="contract-enforcement"></a>Contract Enforcement</h3></div></div><p>
The property-based crosscutting mechanisms can be very useful in
defining more sophisticated contract enforcement. One very powerful
use of these mechanisms is to identify method calls that, in a
correct program, should not exist. For example, the following
aspect enforces the constraint that only the well-known factory
methods can add an element to the registry of figure
elements. Enforcing this constraint ensures that no figure element
is added to the registry more than once.
</p><pre class="programlisting">
aspect RegistrationProtection {
pointcut register(): call(void Registry.register(FigureElement));
pointcut canRegister(): withincode(static * FigureElement.make*(..));
before(): register() &amp;&amp; !canRegister() {
throw new IllegalAccessException("Illegal call " + thisJoinPoint);
}
}
</pre><p>
This aspect uses the withincode primitive pointcut to denote all
join points that occur within the body of the factory methods on
<tt>FigureElement</tt> (the methods with names that
begin with "<tt>make</tt>"). This is a property-based
pointcut because it identifies join points based not on their
signature, but rather on the property that they occur specifically
within the code of another method. The before advice declaration
effectively says signal an error for any calls to register that are
not within the factory methods.
</p><p>
This advice throws a runtime exception at certain join points, but
AspectJ can do better. Using the <tt>declare error</tt>
form, we can have the <span class="emphasis"><i>compiler</i></span> signal the
error.
</p><pre class="programlisting">
aspect RegistrationProtection {
pointcut register(): call(void Registry.register(FigureElement));
pointcut canRegister(): withincode(static * FigureElement.make*(..));
declare error: register() &amp;&amp; !canRegister(): "Illegal call"
}
</pre><p>
When using this aspect, it is impossible for the compiler to
compile programs with these illegal calls. This early detection is
not always possible. In this case, since we depend only on static
information (the <tt>withincode</tt> pointcut picks out
join points totally based on their code, and the
<tt>call</tt> pointcut here picks out join points
statically). Other enforcement, such as the precondition
enforcement, above, does require dynamic information such as the
runtime value of parameters.
</p></div><div class="sect2"><a name="configuration-management"></a><div class="titlepage"><div><h3 class="title"><a name="configuration-management"></a>Configuration Management</h3></div></div><p>
Configuration management for aspects can be handled using a variety
of make-file like techniques. To work with optional aspects, the
programmer can simply define their make files to either include the
aspect in the call to the AspectJ compiler or not, as desired.
</p><p>
Developers who want to be certain that no aspects are included in
the production build can do so by configuring their make files so
that they use a traditional Java compiler for production builds. To
make it easy to write such make files, the AspectJ compiler has a
command-line interface that is consistent with ordinary Java
compilers.
</p></div></div><div class="sect1"><a name="starting-production"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="starting-production"></a>Production Aspects</h2></div></div><p>
This section presents examples of aspects that are inherently
intended to be included in the production builds of an application.
Production aspects tend to add functionality to an application
rather than merely adding more visibility of the internals of a
program. Again, we begin with name-based aspects and follow with
property-based aspects. Name-based production aspects tend to
affect only a small number of methods. For this reason, they are a
good next step for projects adopting AspectJ. But even though they
tend to be small and simple, they can often have a significant
effect in terms of making the program easier to understand and
maintain.
</p><div class="sect2"><a name="change-monitoring"></a><div class="titlepage"><div><h3 class="title"><a name="change-monitoring"></a>Change Monitoring</h3></div></div><p>
The first example production aspect shows how one might implement
some simple functionality where it is problematic to try and do it
explicitly. It supports the code that refreshes the display. The
role of the aspect is to maintain a dirty bit indicating whether or
not an object has moved since the last time the display was
refreshed.
</p><p>
Implementing this functionality as an aspect is straightforward.
The <tt>testAndClear</tt> method is called by the
display code to find out whether a figure element has moved
recently. This method returns the current state of the dirty flag
and resets it to false. The pointcut <tt>move</tt>
captures all the method calls that can move a figure element. The
after advice on <tt>move</tt> sets the dirty flag
whenever an object moves.
</p><pre class="programlisting">
aspect MoveTracking {
private static boolean dirty = false;
public static boolean testAndClear() {
boolean result = dirty;
dirty = false;
return result;
}
pointcut move():
call(void FigureElement.setXY(int, int)) ||
call(void Line.setP1(Point)) ||
call(void Line.setP2(Point)) ||
call(void Point.setX(int)) ||
call(void Point.setY(int));
after() returning: move() {
dirty = true;
}
}
</pre><p>
Even this simple example serves to illustrate some of the important
benefits of using AspectJ in production code. Consider implementing
this functionality with ordinary Java: there would likely be a
helper class that contained the <tt>dirty</tt> flag, the
<tt>testAndClear</tt> method, as well as a
<tt>setFlag</tt> method. Each of the methods that could
move a figure element would include a call to the
<tt>setFlag</tt> method. Those calls, or rather the
concept that those calls should happen at each move operation, are
the crosscutting concern in this case.
</p><p>
The AspectJ implementation has several advantages over the standard
implementation:
</p><p>
<span class="emphasis"><i>The structure of the crosscutting concern is captured
explicitly.</i></span> The moves pointcut clearly states all the
methods involved, so the programmer reading the code sees not just
individual calls to <tt>setFlag</tt>, but instead sees
the real structure of the code. The IDE support included with
AspectJ automatically reminds the programmer that this aspect
advises each of the methods involved. The IDE support also
provides commands to jump to the advice from the method and
vice-versa.
</p><p>
<span class="emphasis"><i>Evolution is easier.</i></span> If, for example, the
aspect needs to be revised to record not just that some figure
element moved, but rather to record exactly which figure elements
moved, the change would be entirely local to the aspect. The
pointcut would be updated to expose the object being moved, and the
advice would be updated to record that object. The paper
<i>An Overview of AspectJ</i> (available linked off
of the AspectJ web site -- <a href="http://eclipse.org/aspectj" target="_top">http://eclipse.org/aspectj</a>), presented at ECOOP
2001, presents a detailed discussion of various ways this aspect
could be expected to evolve.
</p><p>
<span class="emphasis"><i>The functionality is easy to plug in and out.</i></span>
Just as with development aspects, production aspects may need to be
removed from the system, either because the functionality is no
longer needed at all, or because it is not needed in certain
configurations of a system. Because the functionality is
modularized in a single aspect this is easy to do.
</p><p>
<span class="emphasis"><i>The implementation is more stable.</i></span> If, for
example, the programmer adds a subclass of
<tt>Line</tt> that overrides the existing methods,
this advice in this aspect will still apply. In the ordinary Java
implementation the programmer would have to remember to add the
call to <tt>setFlag</tt> in the new overriding
method. This benefit is often even more compelling for
property-based aspects (see the section <a href="#starting-production-consistentBehavior">Providing Consistent Behavior</a>).
</p></div><div class="sect2"><a name="context-passing"></a><div class="titlepage"><div><h3 class="title"><a name="context-passing"></a>Context Passing</h3></div></div><p>
The crosscutting structure of context passing can be a significant
source of complexity in Java programs. Consider implementing
functionality that would allow a client of the figure editor (a
program client rather than a human) to set the color of any figure
elements that are created. Typically this requires passing a color,
or a color factory, from the client, down through the calls that
lead to the figure element factory. All programmers are familiar
with the inconvenience of adding a first argument to a number of
methods just to pass this kind of context information.
</p><p>
Using AspectJ, this kind of context passing can be implemented in a
modular way. The following code adds after advice that runs only
when the factory methods of <tt>Figure</tt> are
called in the control flow of a method on a
<tt>ColorControllingClient</tt>.
</p><pre class="programlisting">
aspect ColorControl {
pointcut CCClientCflow(ColorControllingClient client):
cflow(call(* * (..)) &amp;&amp; target(client));
pointcut make(): call(FigureElement Figure.make*(..));
after (ColorControllingClient c) returning (FigureElement fe):
make() &amp;&amp; CCClientCflow(c) {
fe.setColor(c.colorFor(fe));
}
}
</pre><p>
This aspect affects only a small number of methods, but note that
the non-AOP implementation of this functionality might require
editing many more methods, specifically, all the methods in the
control flow from the client to the factory. This is a benefit
common to many property-based aspects while the aspect is short and
affects only a modest number of benefits, the complexity the aspect
saves is potentially much larger.
</p></div><div class="sect2"><a name="starting-production-consistentBehavior"></a><div class="titlepage"><div><h3 class="title"><a name="starting-production-consistentBehavior"></a>Providing Consistent Behavior</h3></div></div><p>
This example shows how a property-based aspect can be used to
provide consistent handling of functionality across a large set of
operations. This aspect ensures that all public methods of the
<tt>com.bigboxco</tt> package log any Errors they throw
to their caller (in Java, an Error is like an Exception, but it
indicates that something really bad and usually unrecoverable has
happened). The <tt>publicMethodCall</tt> pointcut
captures the public method calls of the package, and the after
advice runs whenever one of those calls throws an Error. The advice
logs that Error and then the throw resumes.
</p><pre class="programlisting">
aspect PublicErrorLogging {
Log log = new Log();
pointcut publicMethodCall():
call(public * com.bigboxco.*.*(..));
after() throwing (Error e): publicMethodCall() {
log.write(e);
}
}
</pre><p>
In some cases this aspect can log an exception twice. This happens
if code inside the <tt>com.bigboxco</tt> package itself
calls a public method of the package. In that case this code will
log the error at both the outermost call into the
<tt>com.bigboxco</tt> package and the re-entrant
call. The <tt>cflow</tt> primitive pointcut can be used
in a nice way to exclude these re-entrant calls:</p><pre class="programlisting">
after() throwing (Error e):
publicMethodCall() &amp;&amp; !cflow(publicMethodCall()) {
log.write(e);
}
</pre><p>
The following aspect is taken from work on the AspectJ compiler.
The aspect advises about 35 methods in the
<tt>JavaParser</tt> class. The individual methods
handle each of the different kinds of elements that must be
parsed. They have names like <tt>parseMethodDec</tt>,
<tt>parseThrows</tt>, and
<tt>parseExpr</tt>.
</p><pre class="programlisting">
aspect ContextFilling {
pointcut parse(JavaParser jp):
call(* JavaParser.parse*(..))
&amp;&amp; target(jp)
&amp;&amp; !call(Stmt parseVarDec(boolean)); // var decs
// are tricky
around(JavaParser jp) returns ASTObject: parse(jp) {
Token beginToken = jp.peekToken();
ASTObject ret = proceed(jp);
if (ret != null) jp.addContext(ret, beginToken);
return ret;
}
}
</pre><p>
This example exhibits a property found in many aspects with large
property-based pointcuts. In addition to a general property based
pattern <tt>call(* JavaParser.parse*(..))</tt> it
includes an exception to the pattern <tt>!call(Stmt
parseVarDec(boolean))</tt>. The exclusion of
<tt>parseVarDec</tt> happens because the parsing of
variable declarations in Java is too complex to fit with the clean
pattern of the other <tt>parse*</tt> methods. Even with
the explicit exclusion this aspect is a clear expression of a clean
crosscutting modularity. Namely that all
<tt>parse*</tt> methods that return
<tt>ASTObjects</tt>, except for
<tt>parseVarDec</tt> share a common behavior for
establishing the parse context of their result.
</p><p>
The process of writing an aspect with a large property-based
pointcut, and of developing the appropriate exceptions can clarify
the structure of the system. This is especially true, as in this
case, when refactoring existing code to use aspects. When we first
looked at the code for this aspect, we were able to use the IDE
support provided in AJDE for JBuilder to see what methods the
aspect was advising compared to our manual coding. We quickly
discovered that there were a dozen places where the aspect advice
was in effect but we had not manually inserted the required
functionality. Two of these were bugs in our prior non-AOP
implementation of the parser. The other ten were needless
performance optimizations. So, here, refactoring the code to
express the crosscutting structure of the aspect explicitly made
the code more concise and eliminated latent bugs.
</p></div></div><div class="sect1"><a name="starting-conclusion"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="starting-conclusion"></a>Conclusion</h2></div></div><p>
AspectJ is a simple and practical aspect-oriented extension to
Java. With just a few new constructs, AspectJ provides support for
modular implementation of a range of crosscutting concerns.
</p><p>
Adoption of AspectJ into an existing Java development project can be
a straightforward and incremental task. One path is to begin by using
only development aspects, going on to using production aspects and
then reusable aspects after building up experience with
AspectJ. Adoption can follow other paths as well. For example, some
developers will benefit from using production aspects right
away. Others may be able to write clean reusable aspects almost right
away.
</p><p>
AspectJ enables both name-based and property based crosscutting.
Aspects that use name-based crosscutting tend to affect a small
number of other classes. But despite their small scale, they can
often eliminate significant complexity compared to an ordinary Java
implementation. Aspects that use property-based crosscutting can
have small or large scale.
</p><p>
Using AspectJ results in clean well-modularized implementations of
crosscutting concerns. When written as an AspectJ aspect the
structure of a crosscutting concern is explicit and easy to
understand. Aspects are also highly modular, making it possible to
develop plug-and-play implementations of crosscutting
functionality.
</p><p>
AspectJ provides more functionality than was covered by this short
introduction. The next chapter, <a href="#language">The AspectJ Language</a>,
covers in detail more of the features of the AspectJ language. The
following chapter, <a href="#examples">Examples</a>, then presents some
carefully chosen examples that show you how AspectJ might be used. We
recommend that you read the next two chapters carefully before
deciding to adopt AspectJ into a project.
</p></div></div><div class="chapter"><div class="titlepage"><div><h2 class="title"><a name="language"></a>Chapter 2. The AspectJ Language</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#language-intro">Introduction</a></dt><dt><a href="#language-anatomy">The Anatomy of an Aspect</a></dt><dd><dl><dt><a href="#an-example-aspect">An Example Aspect</a></dt><dt><a href="#pointcuts">Pointcuts</a></dt><dt><a href="#advice">Advice</a></dt></dl></dd><dt><a href="#language-joinPoints">Join Points and Pointcuts</a></dt><dd><dl><dt><a href="#some-example-pointcuts">Some Example Pointcuts</a></dt><dt><a href="#call-vs-execution">call vs. execution</a></dt><dt><a href="#pointcut-composition">Pointcut composition</a></dt><dt><a href="#pointcut-parameters">Pointcut Parameters</a></dt><dt><a href="#example">Example: <tt>HandleLiveness</tt></a></dt></dl></dd><dt><a href="#language-advice">Advice</a></dt><dt><a href="#language-interType">Inter-type declarations</a></dt><dd><dl><dt><a href="#inter-type-scope">Inter-type Scope</a></dt><dt><a href="#example-pointassertions">Example: <tt>PointAssertions</tt></a></dt></dl></dd><dt><a href="#language-thisJoinPoint">thisJoinPoint</a></dt></dl></div><div class="sect1"><a name="language-intro"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="language-intro"></a>Introduction</h2></div></div><p>
The previous chapter, <a href="#starting">Getting Started with AspectJ</a>, was a brief
overview of the AspectJ language. You should read this chapter to
understand AspectJ's syntax and semantics. It covers the same
material as the previous chapter, but more completely and in much
more detail.
</p><p>
We will start out by looking at an example aspect that we'll build
out of a pointcut, an introduction, and two pieces of advice. This
example aspect will gives us something concrete to talk about.
</p></div><div class="sect1"><a name="language-anatomy"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="language-anatomy"></a>The Anatomy of an Aspect</h2></div></div><p>
This lesson explains the parts of AspectJ's aspects. By reading this
lesson you will have an overview of what's in an aspect and you will
be exposed to the new terminology introduced by AspectJ.
</p><div class="sect2"><a name="an-example-aspect"></a><div class="titlepage"><div><h3 class="title"><a name="an-example-aspect"></a>An Example Aspect</h3></div></div><p>
Here's an example of an aspect definition in AspectJ:
</p><pre class="programlisting">
1 aspect FaultHandler {
2
3 private boolean Server.disabled = false;
4
5 private void reportFault() {
6 System.out.println("Failure! Please fix it.");
7 }
8
9 public static void fixServer(Server s) {
10 s.disabled = false;
11 }
12
13 pointcut services(Server s): target(s) &amp;&amp; call(public * *(..));
14
15 before(Server s): services(s) {
16 if (s.disabled) throw new DisabledException();
17 }
18
19 after(Server s) throwing (FaultException e): services(s) {
20 s.disabled = true;
21 reportFault();
22 }
23 }
</pre><p>
The <tt>FaultHandler</tt> consists of one inter-type
field on <tt>Server</tt> (line 03), two methods (lines
05-07 and 09-11), one pointcut definition (line 13), and two pieces
of advice (lines 15-17 and 19-22).
</p><p>
This covers the basics of what aspects can contain. In general,
aspects consist of an association of other program entities,
ordinary variables and methods, pointcut definitions, inter-type declarations,
and advice, where advice may be before, after or around advice. The
remainder of this lesson focuses on those crosscut-related
constructs.
</p></div><div class="sect2"><a name="pointcuts"></a><div class="titlepage"><div><h3 class="title"><a name="pointcuts"></a>Pointcuts</h3></div></div><p>
AspectJ's pointcut definitions give names to pointcuts. Pointcuts
themselves pick out join points, i.e. interesting points in the
execution of a program. These join points can be method or
constructor invocations and executions, the handling of exceptions,
field assignments and accesses, etc. Take, for example, the
pointcut definition in line 13:
</p><pre class="programlisting">
pointcut services(Server s): target(s) &amp;&amp; call(public * *(..))
</pre><p>
This pointcut, named <tt>services</tt>, picks out those
points in the execution of the program when
<tt>Server</tt> objects have their public methods called.
It also allows anyone using the <tt>services</tt>
pointcut to access the <tt>Server</tt> object whose
method is being called.
</p><p>
The idea behind this pointcut in the
<tt>FaultHandler</tt> aspect is that
fault-handling-related behavior must be triggered on the calls to
public methods. For example, the server may be unable to proceed
with the request because of some fault. The calls of those methods
are, therefore, interesting events for this aspect, in the sense
that certain fault-related things will happen when these events
occur.
</p><p>
Part of the context in which the events occur is exposed by the
formal parameters of the pointcut. In this case, that consists of
objects of type <tt>Server</tt>. That formal parameter
is then being used on the right hand side of the declaration in
order to identify which events the pointcut refers to. In this
case, a pointcut picking out join points where a Server is the
target of some operation (target(s)) is being composed
(<tt>&amp;&amp;</tt>, meaning and) with a pointcut
picking out call join points (call(...)). The calls are identified
by signatures that can include wild cards. In this case, there are
wild cards in the return type position (first *), in the name
position (second *) and in the argument list position (..); the
only concrete information is given by the qualifier
<tt>public</tt>.
</p><p>
Pointcuts pick out arbitrarily large numbers of join points of a
program. But they pick out only a small number of
<span class="emphasis"><i>kinds</i></span> of join points. Those kinds of join
points correspond to some of the most important concepts in
Java. Here is an incomplete list: method call, method execution,
exception handling, instantiation, constructor execution, and
field access. Each kind of join point can be picked out by its
own specialized pointcut that you will learn about in other parts
of this guide.
</p></div><div class="sect2"><a name="advice"></a><div class="titlepage"><div><h3 class="title"><a name="advice"></a>Advice</h3></div></div><p>
A piece of advice brings together a pointcut and a body of code to
define aspect implementation that runs at join points picked out by
the pointcut. For example, the advice in lines 15-17 specifies that
the following piece of code
</p><pre class="programlisting">
{
if (s.disabled) throw new DisabledException();
}
</pre><p>
is executed when instances of the <tt>Server</tt> class
have their public methods called, as specified by the pointcut
<tt>services</tt>. More specifically, it runs when those
calls are made, just before the corresponding methods are executed.
</p><p>
The advice in lines 19-22 defines another piece of implementation
that is executed on the same pointcut:
</p><pre class="programlisting">
{
s.disabled = true;
reportFault();
}
</pre><p>
But this second method executes after those operations throw
exception of type <tt>FaultException</tt>.
</p><p>
There are two other variations of after advice: upon successful
return and upon return, either successful or with an exception.
There is also a third kind of advice called around. You will see
those in other parts of this guide.
</p></div></div><div class="sect1"><a name="language-joinPoints"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="language-joinPoints"></a>Join Points and Pointcuts</h2></div></div><p>
Consider the following Java class:
</p><pre class="programlisting">
class Point {
private int x, y;
Point(int x, int y) { this.x = x; this.y = y; }
void setX(int x) { this.x = x; }
void setY(int y) { this.y = y; }
int getX() { return x; }
int getY() { return y; }
}
</pre><p>
In order to get an intuitive understanding of AspectJ's join points
and pointcuts, let's go back to some of the basic principles of
Java. Consider the following a method declaration in class Point:
</p><pre class="programlisting">
void setX(int x) { this.x = x; }
</pre><p>
This piece of program says that that when method named
<tt>setX</tt> with an <tt>int</tt> argument
called on an object of type <tt>Point</tt>, then the method
body <tt>{ this.x = x; }</tt> is executed. Similarly, the
constructor of the class states that when an object of type
<tt>Point</tt> is instantiated through a constructor with
two <tt>int</tt> arguments, then the constructor body
<tt>{ this.x = x; this.y = y; }</tt> is executed.
</p><p>
One pattern that emerges from these descriptions is
<blockquote class="blockquote">
When something happens, then something gets executed.
</blockquote>
In object-oriented programs, there are several kinds of "things that
happen" that are determined by the language. We call these the join
points of Java. Join points consist of things like method calls,
method executions, object instantiations, constructor executions,
field references and handler executions. (See the <a href="#quick">AspectJ Quick Reference</a> for a complete listing.)
</p><p>
Pointcuts pick out these join points. For example, the pointcut
</p><pre class="programlisting">
pointcut setter(): target(Point) &amp;&amp;
(call(void setX(int)) ||
call(void setY(int)));
</pre><p>
picks out each call to <tt>setX(int)</tt> or
<tt>setY(int)</tt> when called on an instance of
<tt>Point</tt>. Here's another example:
</p><pre class="programlisting">
pointcut ioHandler(): within(MyClass) &amp;&amp; handler(IOException);
</pre><p>
This pointcut picks out each the join point when exceptions of type
<tt>IOException</tt> are handled inside the code defined by
class <tt>MyClass</tt>.
</p><p>
Pointcut definitions consist of a left-hand side and a right-hand side,
separated by a colon. The left-hand side consists of the pointcut name
and the pointcut parameters (i.e. the data available when the events
happen). The right-hand side consists of the pointcut itself.
</p><div class="sect2"><a name="some-example-pointcuts"></a><div class="titlepage"><div><h3 class="title"><a name="some-example-pointcuts"></a>Some Example Pointcuts</h3></div></div><p>
Here are examples of pointcuts picking out
</p><div class="variablelist"><dl><dt><a name="d0e1024"></a><span class="term">when a particular method body executes</span></dt><dd><p><a name="d0e1027"></a>
<tt>execution(void Point.setX(int))</tt>
</p></dd><dt><a name="d0e1033"></a><span class="term">when a method is called</span></dt><dd><p><a name="d0e1036"></a>
<tt>call(void Point.setX(int))</tt>
</p></dd><dt><a name="d0e1042"></a><span class="term">when an exception handler executes</span></dt><dd><p><a name="d0e1045"></a>
<tt>handler(ArrayOutOfBoundsException)</tt>
</p></dd><dt><a name="d0e1051"></a><span class="term">
when the object currently executing
(i.e. <tt>this</tt>) is of type
<tt>SomeType</tt>
</span></dt><dd><p><a name="d0e1060"></a>
<tt>this(SomeType)</tt>
</p></dd><dt><a name="d0e1066"></a><span class="term">
when the target object is of type <tt>SomeType</tt>
</span></dt><dd><p><a name="d0e1072"></a>
<tt>target(SomeType)</tt>
</p></dd><dt><a name="d0e1078"></a><span class="term">
when the executing code belongs to
class <tt>MyClass</tt>
</span></dt><dd><p><a name="d0e1084"></a>
<tt>within(MyClass)</tt>
</p></dd><dt><a name="d0e1090"></a><span class="term">
when the join point is in the control flow of a call to a
<tt>Test</tt>'s no-argument <tt>main</tt>
method
</span></dt><dd><p><a name="d0e1099"></a>
<tt>cflow(call(void Test.main()))</tt>
</p></dd></dl></div><p>
Pointcuts compose through the operations <tt>or</tt>
("<tt>||</tt>"), <tt>and</tt>
("<tt>&amp;&amp;</tt>") and <tt>not</tt>
("<tt>!</tt>").
</p><div class="itemizedlist"><ul><li><p><a name="d0e1126"></a>
It is possible to use wildcards. So
<div class="orderedlist"><ol type="1"><li><p><a name="d0e1130"></a>
<tt>execution(* *(..))</tt>
</p></li><li><p><a name="d0e1136"></a>
<tt>call(* set(..))</tt>
</p></li></ol></div>
means (1) the execution of any method regardless of return or
parameter types, and (2) the call to any method named
<tt>set</tt> regardless of return or parameter types
-- in case of overloading there may be more than one such
<tt>set</tt> method; this pointcut picks out calls to
all of them.
</p></li><li><p><a name="d0e1149"></a>
You can select elements based on types. For example,
<div class="orderedlist"><ol type="1"><li><p><a name="d0e1153"></a>
<tt>execution(int *())</tt>
</p></li><li><p><a name="d0e1159"></a>
<tt>call(* setY(long))</tt>
</p></li><li><p><a name="d0e1165"></a>
<tt>call(* Point.setY(int))</tt>
</p></li><li><p><a name="d0e1171"></a>
<tt>call(*.new(int, int))</tt>
</p></li></ol></div>
means (1) the execution of any method with no parameters that
returns an <tt>int</tt>, (2) the call to any
<tt>setY</tt> method that takes a
<tt>long</tt> as an argument, regardless of return
type or declaring type, (3) the call to any of
<tt>Point</tt>'s <tt>setY</tt> methods that
take an <tt>int</tt> as an argument, regardless of
return type, and (4) the call to any classes' constructor, so
long as it takes exactly two <tt>int</tt>s as
arguments.
</p></li><li><p><a name="d0e1199"></a>
You can compose pointcuts. For example,
<div class="orderedlist"><ol type="1"><li><p><a name="d0e1203"></a>
<tt>target(Point) &amp;&amp; call(int *())</tt>
</p></li><li><p><a name="d0e1209"></a>
<tt>call(* *(..)) &amp;&amp; (within(Line) || within(Point))</tt>
</p></li><li><p><a name="d0e1215"></a>
<tt>within(*) &amp;&amp; execution(*.new(int))</tt>
</p></li><li><p><a name="d0e1221"></a>
<tt>
!this(Point) &amp;&amp; call(int *(..))
</tt>
</p></li></ol></div>
means (1) any call to an <tt>int</tt> method with no
arguments on an instance of <tt>Point</tt>,
regardless of its name, (2) any call to any method where the
call is made from the code in <tt>Point</tt>'s or
<tt>Line</tt>'s type declaration, (3) the execution of
any constructor taking exactly one <tt>int</tt>
argument, regardless of where the call is made from, and
(4) any method call to an <tt>int</tt> method when
the executing object is any type except <tt>Point</tt>.
</p></li><li><p><a name="d0e1249"></a>
You can select methods and constructors based on their
modifiers and on negations of modifiers. For example, you can
say:
<div class="orderedlist"><ol type="1"><li><p><a name="d0e1253"></a>
<tt>call(public * *(..))</tt>
</p></li><li><p><a name="d0e1259"></a>
<tt>execution(!static * *(..))</tt>
</p></li><li><p><a name="d0e1265"></a>
<tt> execution(public !static * *(..))</tt>
</p></li></ol></div>
which means (1) any call to a public method, (2) any
execution of a non-static method, and (3) any execution of a
public, non-static method.
</p></li><li><p><a name="d0e1272"></a>
Pointcuts can also deal with interfaces. For example, given the
interface </p><pre class="programlisting">
interface MyInterface { ... }
</pre><p>
the pointcut <tt>call(* MyInterface.*(..))</tt> picks
out any call to a method in <tt>MyInterface</tt>'s
signature -- that is, any method defined by
<tt>MyInterface</tt> or inherited by one of its a
supertypes.
</p></li></ul></div></div><div class="sect2"><a name="call-vs-execution"></a><div class="titlepage"><div><h3 class="title"><a name="call-vs-execution"></a>call vs. execution</h3></div></div><p>
When methods and constructors run, there are two interesting times
associated with them. That is when they are called, and when they
actually execute.
</p><p>
AspectJ exposes these times as call and execution join points,
respectively, and allows them to be picked out specifically by
<tt>call</tt> and <tt>execution</tt> pointcuts.
</p><p>
So what's the difference between these join points? Well, there are a
number of differences:
</p><p>
Firstly, the lexical pointcut declarations
<tt>within</tt> and <tt>withincode</tt> match
differently. At a call join point, the enclosing code is that of
the call site. This means that <tt>call(void m())
&amp;&amp; withincode(void m())</tt> will only capture
directly recursive calls, for example. At an execution join point,
however, the program is already executing the method, so the
enclosing code is the method itself: <tt>execution(void m())
&amp;&amp; withincode(void m())</tt> is the same as
<tt>execution(void m())</tt>.
</p><p>
Secondly, the call join point does not capture super calls to
non-static methods. This is because such super calls are different in
Java, since they don't behave via dynamic dispatch like other calls to
non-static methods.
</p><p>
The rule of thumb is that if you want to pick a join point that
runs when an actual piece of code runs (as is often the case for
tracing), use <tt>execution</tt>, but if you want to pick
one that runs when a particular <span class="emphasis"><i>signature</i></span> is
called (as is often the case for production aspects), use
<tt>call</tt>.
</p></div><div class="sect2"><a name="pointcut-composition"></a><div class="titlepage"><div><h3 class="title"><a name="pointcut-composition"></a>Pointcut composition</h3></div></div><p>
Pointcuts are put together with the operators and (spelled
<tt>&amp;&amp;</tt>), or (spelled <tt>||</tt>),
and not (spelled <tt>!</tt>). This allows the creation
of very powerful pointcuts from the simple building blocks of
primitive pointcuts. This composition can be somewhat confusing
when used with primitive pointcuts like <tt>cflow</tt>
and <tt>cflowbelow</tt>. Here's an example:
</p><p>
<tt>cflow(<i><tt>P</tt></i>)</tt> picks out
each join point in the control flow of the join points picked out
by <i><tt>P</tt></i>. So, pictorially:
</p><pre class="programlisting">
P ---------------------
\
\ cflow of P
\
</pre><p>
What does <tt>cflow(<i><tt>P</tt></i>) &amp;&amp;
cflow(<i><tt>Q</tt></i>)</tt> pick out? Well, it
picks out each join point that is in both the control flow of
<i><tt>P</tt></i> and in the control flow of
<i><tt>Q</tt></i>. So...
</p><pre class="programlisting">
P ---------------------
\
\ cflow of P
\
\
\
Q -------------\-------
\ \
\ cflow of Q \ cflow(P) &amp;&amp; cflow(Q)
\ \
</pre><p>
Note that <i><tt>P</tt></i> and
<i><tt>Q</tt></i> might not have any join points in
common... but their control flows might have join points in common.
</p><p>
But what does <tt>cflow(<i><tt>P</tt></i>
&amp;&amp; <i><tt>Q</tt></i>)</tt> mean? Well, it
means the control flow of those join points that are both picked
out by <i><tt>P</tt></i> and picked out by
<i><tt>Q</tt></i>.
</p><pre class="programlisting">
P &amp;&amp; Q -------------------
\
\ cflow of (P &amp;&amp; Q)
\
</pre><p>
and if there are <span class="emphasis"><i>no</i></span> join points that are both
picked by <i><tt>P</tt></i> and picked out by
<i><tt>Q</tt></i>, then there's no chance that there are
any join points in the control flow of
<tt>(<i><tt>P</tt></i> &amp;&amp;
<i><tt>Q</tt></i>)</tt>.
</p><p>
Here's some code that expresses this.
</p><pre class="programlisting">
public class Test {
public static void main(String[] args) {
foo();
}
static void foo() {
goo();
}
static void goo() {
System.out.println("hi");
}
}
aspect A {
pointcut fooPC(): execution(void Test.foo());
pointcut gooPC(): execution(void Test.goo());
pointcut printPC(): call(void java.io.PrintStream.println(String));
before(): cflow(fooPC()) &amp;&amp; cflow(gooPC()) &amp;&amp; printPC() &amp;&amp; !within(A) {
System.out.println("should occur");
}
before(): cflow(fooPC() &amp;&amp; gooPC()) &amp;&amp; printPC() &amp;&amp; !within(A) {
System.out.println("should not occur");
}
}
</pre><p>
The <tt>!within(<i><tt>A</tt></i>)</tt>
pointcut above is required to avoid the <tt>printPC</tt>
pointcut applying to the <tt>System.out.println</tt>
call in the advice body. If this was not present a recursive call
would result as the pointcut would apply to it's own advice.
(See <a href="#pitfalls-infiniteLoops" title="Infinite loops">the section called &#8220;Infinite loops&#8221;</a> for more details.)
</p></div><div class="sect2"><a name="pointcut-parameters"></a><div class="titlepage"><div><h3 class="title"><a name="pointcut-parameters"></a>Pointcut Parameters</h3></div></div><p>
Consider again the first pointcut definition in this chapter:
</p><pre class="programlisting">
pointcut setter(): target(Point) &amp;&amp;
(call(void setX(int)) ||
call(void setY(int)));
</pre><p>
As we've seen, this pointcut picks out each call to
<tt>setX(int)</tt> or <tt>setY(int)</tt>
methods where the target is an instance of
<tt>Point</tt>. The pointcut is given the name
<tt>setters</tt> and no parameters on the left-hand
side. An empty parameter list means that none of the context from
the join points is published from this pointcut. But consider
another version of version of this pointcut definition:
</p><pre class="programlisting">
pointcut setter(Point p): target(p) &amp;&amp;
(call(void setX(int)) ||
call(void setY(int)));
</pre><p>
This version picks out exactly the same join points. But in this
version, the pointcut has one parameter of type
<tt>Point</tt>. This means that any advice that uses this
pointcut has access to a <tt>Point</tt> from each join
point picked out by the pointcut. Inside the pointcut definition
this <tt>Point</tt> is named <tt>p</tt> is
available, and according to the right-hand side of the definition,
that <tt>Point p</tt> comes from the
<tt>target</tt> of each matched join point.
</p><p>
Here's another example that illustrates the flexible mechanism for
defining pointcut parameters:
</p><pre class="programlisting">
pointcut testEquality(Point p): target(Point) &amp;&amp;
args(p) &amp;&amp;
call(boolean equals(Object));
</pre><p>
This pointcut also has a parameter of type
<tt>Point</tt>. Similar to the
<tt>setters</tt> pointcut, this means that anyone using
this pointcut has access to a <tt>Point</tt> from each
join point. But in this case, looking at the right-hand side we
find that the object named in the parameters is not the target
<tt>Point</tt> object that receives the call; it's the
argument (also of type <tt>Point</tt>) passed to the
<tt>equals</tt> method when some other
<tt>Point</tt> is the target. If we wanted access to both
<tt>Point</tt>s, then the pointcut definition that would
expose target <tt>Point p1</tt> and argument
<tt>Point p2</tt> would be
</p><pre class="programlisting">
pointcut testEquality(Point p1, Point p2): target(p1) &amp;&amp;
args(p2) &amp;&amp;
call(boolean equals(Object));
</pre><p>
Let's look at another variation of the <tt>setters</tt> pointcut:
</p><pre class="programlisting">
pointcut setter(Point p, int newval): target(p) &amp;&amp;
args(newval) &amp;&amp;
(call(void setX(int)) ||
call(void setY(int)));
</pre><p>
In this case, a <tt>Point</tt> object and an
<tt>int</tt> value are exposed by the named
pointcut. Looking at the the right-hand side of the definition, we
find that the <tt>Point</tt> object is the target object,
and the <tt>int</tt> value is the called method's
argument.
</p><p>
The use of pointcut parameters is relatively flexible. The most
important rule is that all the pointcut parameters must be bound at
every join point picked out by the pointcut. So, for example, the
following pointcut definition will result in a compilation error:
<pre class="programlisting">
pointcut badPointcut(Point p1, Point p2):
(target(p1) &amp;&amp; call(void setX(int))) ||
(target(p2) &amp;&amp; call(void setY(int)));
</pre>
because <tt>p1</tt> is only bound when calling
<tt>setX</tt>, and <tt>p2</tt> is only bound
when calling <tt>setY</tt>, but the pointcut picks out
all of these join points and tries to bind both
<tt>p1</tt> and <tt>p2</tt>.
</p></div><div class="sect2"><a name="example"></a><div class="titlepage"><div><h3 class="title"><a name="example"></a>Example: <tt>HandleLiveness</tt></h3></div></div><p>
The example below consists of two object classes (plus an exception
class) and one aspect. Handle objects delegate their public,
non-static operations to their <tt>Partner</tt>
objects. The aspect <tt>HandleLiveness</tt> ensures that,
before the delegations, the partner exists and is alive, or else it
throws an exception.
</p><pre class="programlisting">
class Handle {
Partner partner = new Partner();
public void foo() { partner.foo(); }
public void bar(int x) { partner.bar(x); }
public static void main(String[] args) {
Handle h1 = new Handle();
h1.foo();
h1.bar(2);
}
}
class Partner {
boolean isAlive() { return true; }
void foo() { System.out.println("foo"); }
void bar(int x) { System.out.println("bar " + x); }
}
aspect HandleLiveness {
before(Handle handle): target(handle) &amp;&amp; call(public * *(..)) {
if ( handle.partner == null || !handle.partner.isAlive() ) {
throw new DeadPartnerException();
}
}
}
class DeadPartnerException extends RuntimeException {}
</pre></div></div><div class="sect1"><a name="language-advice"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="language-advice"></a>Advice</h2></div></div><p>
Advice defines pieces of aspect implementation that execute at
well-defined points in the execution of the program. Those points can
be given either by named pointcuts (like the ones you've seen above)
or by anonymous pointcuts. Here is an example of an advice on a named
pointcut:
</p><pre class="programlisting">
pointcut setter(Point p1, int newval): target(p1) &amp;&amp; args(newval)
(call(void setX(int) ||
call(void setY(int)));
before(Point p1, int newval): setter(p1, newval) {
System.out.println("About to set something in " + p1 +
" to the new value " + newval);
}
</pre><p>
And here is exactly the same example, but using an anonymous
pointcut:
</p><pre class="programlisting">
before(Point p1, int newval): target(p1) &amp;&amp; args(newval)
(call(void setX(int)) ||
call(void setY(int))) {
System.out.println("About to set something in " + p1 +
" to the new value " + newval);
}
</pre><p>
Here are examples of the different advice:
</p><p>
This before advice runs just before the join points picked out by the
(anonymous) pointcut:
</p><pre class="programlisting">
before(Point p, int x): target(p) &amp;&amp; args(x) &amp;&amp; call(void setX(int)) {
if (!p.assertX(x)) return;
}
</pre><p>
This after advice runs just after each join point picked out by the
(anonymous) pointcut, regardless of whether it returns normally or throws
an exception:
</p><pre class="programlisting">
after(Point p, int x): target(p) &amp;&amp; args(x) &amp;&amp; call(void setX(int)) {
if (!p.assertX(x)) throw new PostConditionViolation();
}
</pre><p>
This after returning advice runs just after each join point picked
out by the (anonymous) pointcut, but only if it returns normally.
The return value can be accessed, and is named <tt>x</tt>
here. After the advice runs, the return value is returned:
</p><pre class="programlisting">
after(Point p) returning(int x): target(p) &amp;&amp; call(int getX()) {
System.out.println("Returning int value " + x + " for p = " + p);
}
</pre><p>
This after throwing advice runs just after each join point picked out by
the (anonymous) pointcut, but only when it throws an exception of type
<tt>Exception</tt>. Here the exception value can be accessed
with the name <tt>e</tt>. The advice re-raises the exception
after it's done:
</p><pre class="programlisting">
after() throwing(Exception e): target(Point) &amp;&amp; call(void setX(int)) {
System.out.println(e);
}
</pre><p>
This around advice traps the execution of the join point; it runs
<span class="emphasis"><i>instead</i></span> of the join point. The original action
associated with the join point can be invoked through the special
<tt>proceed</tt> call:
</p><pre class="programlisting">
void around(Point p, int x): target(p)
&amp;&amp; args(x)
&amp;&amp; call(void setX(int)) {
if (p.assertX(x)) proceed(p, x);
p.releaseResources();
}
</pre></div><div class="sect1"><a name="language-interType"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="language-interType"></a>Inter-type declarations</h2></div></div><p>
Aspects can declare members (fields, methods, and constructors) that
are owned by other types. These are called inter-type members.
Aspects can also declare that other types implement new interfaces or
extend a new class. Here are examples of some such inter-type
declarations:
</p><p>
This declares that each <tt>Server</tt> has a
<tt>boolean</tt> field named <tt>disabled</tt>,
initialized to <tt>false</tt>:
<pre class="programlisting">
private boolean Server.disabled = false;
</pre>
It is declared <tt>private</tt>, which means that it is
private <span class="emphasis"><i>to the aspect</i></span>: only code in the aspect
can see the field. And even if <tt>Server</tt> has
another private field named <tt>disabled</tt> (declared in
<tt>Server</tt> or in another aspect) there won't be a name
collision, since no reference to <tt>disabled</tt> will be
ambiguous.
</p><p>
This declares that each <tt>Point</tt> has an
<tt>int</tt> method named <tt>getX</tt> with no
arguments that returns whatever <tt>this.x</tt> is:
<pre class="programlisting">
public int Point.getX() { return this.x; }
</pre>
Inside the body, <tt>this</tt> is the
<tt>Point</tt> object currently executing. Because the
method is publically declared any code can call it, but if there is
some other <tt>Point.getX()</tt> declared there will be a
compile-time conflict.
</p><p>
This publically declares a two-argument constructor for
<tt>Point</tt>:
<pre class="programlisting">
public Point.new(int x, int y) { this.x = x; this.y = y; }
</pre>
</p><p>
This publicly declares that each <tt>Point</tt> has an
<tt>int</tt> field named <tt>x</tt>, initialized
to zero:
<pre class="programlisting">
public int Point.x = 0;
</pre>
Because this is publically declared, it is an error if
<tt>Point</tt> already has a field named
<tt>x</tt> (defined by <tt>Point</tt> or by
another aspect).
</p><p>
This declares that the <tt>Point</tt> class implements the
<tt>Comparable</tt> interface:
<pre class="programlisting">
declare parents: Point implements Comparable;
</pre>
Of course, this will be an error unless <tt>Point</tt>
defines the methods required by <tt>Comparable</tt>.
</p><p>
This declares that the <tt>Point</tt> class extends the
<tt>GeometricObject</tt> class.
<pre class="programlisting">
declare parents: Point extends GeometricObject;
</pre>
</p><p>
An aspect can have several inter-type declarations. For example, the
following declarations
<pre class="programlisting">
public String Point.name;
public void Point.setName(String name) { this.name = name; }
</pre>
publicly declare that Point has both a String field
<tt>name</tt> and a <tt>void</tt> method
<tt>setName(String)</tt> (which refers to the
<tt>name</tt> field declared by the aspect).
</p><p>
An inter-type member can only have one target type, but often you may
wish to declare the same member on more than one type. This can be
done by using an inter-type member in combination with a private
interface:
<pre class="programlisting">
aspect A {
private interface HasName {}
declare parents: (Point || Line || Square) implements HasName;
private String HasName.name;
public String HasName.getName() { return name; }
}
</pre>
This declares a marker interface <tt>HasName</tt>, and also declares that any
type that is either <tt>Point</tt>,
<tt>Line</tt>, or <tt>Square</tt> implements that
interface. It also privately declares that all <tt>HasName</tt>
object have a <tt>String</tt> field called
<tt>name</tt>, and publically declares that all
<tt>HasName</tt> objects have a <tt>String</tt>
method <tt>getName()</tt> (which refers to the privately
declared <tt>name</tt> field).
</p><p>
As you can see from the above example, an aspect can declare that
interfaces have fields and methods, even non-constant fields and
methods with bodies.
</p><div class="sect2"><a name="inter-type-scope"></a><div class="titlepage"><div><h3 class="title"><a name="inter-type-scope"></a>Inter-type Scope</h3></div></div><p>
AspectJ allows private and package-protected (default) inter-type declarations in
addition to public inter-type declarations. Private means private in
relation to the aspect, not necessarily the target type. So, if an
aspect makes a private inter-type declaration of a field
<pre class="programlisting">
private int Foo.x;
</pre>
Then code in the aspect can refer to <tt>Foo</tt>'s
<tt>x</tt> field, but nobody else can. Similarly, if an
aspect makes a package-protected introduction,
</p><pre class="programlisting">
int Foo.x;
</pre><p>
then everything in the aspect's package (which may or may not be
<tt>Foo</tt>'s package) can access <tt>x</tt>.
</p></div><div class="sect2"><a name="example-pointassertions"></a><div class="titlepage"><div><h3 class="title"><a name="example-pointassertions"></a>Example: <tt>PointAssertions</tt></h3></div></div><p>
The example below consists of one class and one aspect. The aspect
privately declares the assertion methods of
<tt>Point</tt>, <tt>assertX</tt> and
<tt>assertY</tt>. It also guards calls to
<tt>setX</tt> and <tt>setY</tt> with calls to
these assertion methods. The assertion methods are declared
privately because other parts of the program (including the code in
<tt>Point</tt>) have no business accessing the assert
methods. Only the code inside of the aspect can call those
methods.
</p><pre class="programlisting">
class Point {
int x, y;
public void setX(int x) { this.x = x; }
public void setY(int y) { this.y = y; }
public static void main(String[] args) {
Point p = new Point();
p.setX(3); p.setY(333);
}
}
aspect PointAssertions {
private boolean Point.assertX(int x) {
return (x &lt;= 100 &amp;&amp; x &gt;= 0);
}
private boolean Point.assertY(int y) {
return (y &lt;= 100 &amp;&amp; y &gt;= 0);
}
before(Point p, int x): target(p) &amp;&amp; args(x) &amp;&amp; call(void setX(int)) {
if (!p.assertX(x)) {
System.out.println("Illegal value for x"); return;
}
}
before(Point p, int y): target(p) &amp;&amp; args(y) &amp;&amp; call(void setY(int)) {
if (!p.assertY(y)) {
System.out.println("Illegal value for y"); return;
}
}
}
</pre></div></div><div class="sect1"><a name="language-thisJoinPoint"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="language-thisJoinPoint"></a>thisJoinPoint</h2></div></div><p>
AspectJ provides a special reference variable,
<tt>thisJoinPoint</tt>, that contains reflective
information about the current join point for the advice to use. The
<tt>thisJoinPoint</tt> variable can only be used in the
context of advice, just like <tt>this</tt> can only be used
in the context of non-static methods and variable initializers. In
advice, <tt>thisJoinPoint</tt> is an object of type <a href="../api/org/aspectj/lang/JoinPoint.html" target="_top"><tt>org.aspectj.lang.JoinPoint</tt></a>.
</p><p>
One way to use it is simply to print it out. Like all Java objects,
<tt>thisJoinPoint</tt> has a <tt>toString()</tt>
method that makes quick-and-dirty tracing easy:
</p><pre class="programlisting">
class TraceNonStaticMethods {
before(Point p): target(p) &amp;&amp; call(* *(..)) {
System.out.println("Entering " + thisJoinPoint + " in " + p);
}
}
</pre><p>
The type of <tt>thisJoinPoint</tt> includes a rich
reflective class hierarchy of signatures, and can be used to access
both static and dynamic information about join points such as the
arguments of the join point:
<pre class="programlisting">
thisJoinPoint.getArgs()
</pre>
In addition, it holds an object consisting of all the static
information about the join point such as corresponding line number
and static signature:
<pre class="programlisting">
thisJoinPoint.getStaticPart()
</pre>
If you only need the static information about the join point, you may
access the static part of the join point directly with the special
variable <tt>thisJoinPointStaticPart</tt>. Using
<tt>thisJoinPointStaticPart</tt> will avoid the run-time
creation of the join point object that may be necessary when using
<tt>thisJoinPoint</tt> directly.
</p><p>It is always the case that
</p><pre class="programlisting">
thisJoinPointStaticPart == thisJoinPoint.getStaticPart()
thisJoinPoint.getKind() == thisJoinPointStaticPart.getKind()
thisJoinPoint.getSignature() == thisJoinPointStaticPart.getSignature()
thisJoinPoint.getSourceLocation() == thisJoinPointStaticPart.getSourceLocation()
</pre><p>
One more reflective variable is available:
<tt>thisEnclosingJoinPointStaticPart</tt>. This, like
<tt>thisJoinPointStaticPart</tt>, only holds the static
part of a join point, but it is not the current but the enclosing
join point. So, for example, it is possible to print out the calling
source location (if available) with
</p><pre class="programlisting">
before() : execution (* *(..)) {
System.err.println(thisEnclosingJoinPointStaticPart.getSourceLocation())
}
</pre></div></div><div class="chapter"><div class="titlepage"><div><h2 class="title"><a name="examples"></a>Chapter 3. Examples</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#examples-intro">Introduction</a></dt><dt><a href="#examples-howto">Obtaining, Compiling and Running the Examples</a></dt><dt><a href="#examples-basic">Basic Techniques</a></dt><dd><dl><dt><a href="#examples-joinPoints">Join Points and <tt>thisJoinPoint</tt></a></dt><dt><a href="#examples-roles">Roles and Views</a></dt></dl></dd><dt><a href="#examples-development">Development Aspects</a></dt><dd><dl><dt><a href="#tracing-using-aspects">Tracing using aspects</a></dt></dl></dd><dt><a href="#examples-production">Production Aspects</a></dt><dd><dl><dt><a href="#a-bean-aspect">A Bean Aspect</a></dt><dt><a href="#the-subject-observer-protocol">The Subject/Observer Protocol</a></dt><dt><a href="#a-simple-telecom-simulation">A Simple Telecom Simulation</a></dt></dl></dd><dt><a href="#examples-reusable">Reusable Aspects</a></dt><dd><dl><dt><a href="#tracing-using-aspects-revisited">Tracing using Aspects, Revisited</a></dt></dl></dd></dl></div><div class="sect1"><a name="examples-intro"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="examples-intro"></a>Introduction</h2></div></div><p>
This chapter consists entirely of examples of AspectJ use.
</p><p>The examples can be grouped into four categories:</p><table class="simplelist" border="0" summary="Simple list"><tr><td><span class="bold"><b>technique</b></span></td><td>Examples which illustrate how to use one or more features of the
language. </td></tr><tr><td><span class="bold"><b>development</b></span></td><td>Examples of using AspectJ during the development phase of a
project. </td></tr><tr><td><span class="bold"><b>production</b></span></td><td>Examples of using AspectJ to provide functionality in an
application. </td></tr><tr><td><span class="bold"><b>reusable</b></span></td><td>Examples of reuse of aspects and pointcuts.</td></tr></table></div><div class="sect1"><a name="examples-howto"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="examples-howto"></a>Obtaining, Compiling and Running the Examples</h2></div></div><p>
The examples source code is part of the AspectJ distribution which may be
downloaded from the AspectJ project page ( <a href="http://eclipse.org/aspectj" target="_top">http://eclipse.org/aspectj</a> ).
</p><p>
Compiling most examples is straightforward. Go the
<tt><i><tt>InstallDir</tt></i>/examples</tt>
directory, and look for a <tt>.lst</tt> file in one of
the example subdirectories. Use the <tt>-arglist</tt>
option to <tt>ajc</tt> to compile the example. For
instance, to compile the telecom example with billing, type
</p><pre class="programlisting">
ajc -argfile telecom/billing.lst
</pre><p>
To run the examples, your classpath must include the AspectJ run-time
Java archive (<tt>aspectjrt.jar</tt>). You may either set the
<tt>CLASSPATH</tt> environment variable or use the
<tt>-classpath</tt> command line option to the Java
interpreter:
</p><pre class="programlisting">
(In Unix use a : in the CLASSPATH)
java -classpath ".:<i><tt>InstallDir</tt></i>/lib/aspectjrt.jar" telecom.billingSimulation
</pre><pre class="programlisting">
(In Windows use a ; in the CLASSPATH)
java -classpath ".;<i><tt>InstallDir</tt></i>/lib/aspectjrt.jar" telecom.billingSimulation
</pre></div><div class="sect1"><a name="examples-basic"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="examples-basic"></a>Basic Techniques</h2></div></div><p>
This section presents two basic techniques of using AspectJ, one each
from the two fundamental ways of capturing crosscutting concerns:
with dynamic join points and advice, and with static
introduction. Advice changes an application's behavior. Introduction
changes both an application's behavior and its structure.
</p><p>
The first example, <a href="#examples-joinPoints" title="Join Points and thisJoinPoint">the section called &#8220;Join Points and <tt>thisJoinPoint</tt>&#8221;</a>, is about
gathering and using information about the join point that has
triggered some advice. The second example, <a href="#examples-roles" title="Roles and Views">the section called &#8220;Roles and Views&#8221;</a>, concerns a crosscutting view of an
existing class hierarchy. </p><div class="sect2"><a name="examples-joinPoints"></a><div class="titlepage"><div><h3 class="title"><a name="examples-joinPoints"></a>Join Points and <tt>thisJoinPoint</tt></h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/tjp</tt>.)
</p><p>
A join point is some point in the execution of a program together
with a view into the execution context when that point occurs. Join
points are picked out by pointcuts. When a program reaches a join
point, advice on that join point may run in addition to (or instead
of) the join point itself.
</p><p>
When using a pointcut that picks out join points of a single kind
by name, typicaly the the advice will know exactly what kind of
join point it is associated with. The pointcut may even publish
context about the join point. Here, for example, since the only
join points picked out by the pointcut are calls of a certain
method, we can get the target value and one of the argument values
of the method calls directly.
</p><pre class="programlisting">
before(Point p, int x): target(p)
&amp;&amp; args(x)
&amp;&amp; call(void setX(int)) {
if (!p.assertX(x)) {
System.out.println("Illegal value for x"); return;
}
}
</pre><p>
But sometimes the shape of the join point is not so clear. For
instance, suppose a complex application is being debugged, and we
want to trace when any method of some class is executed. The
pointcut
</p><pre class="programlisting">
pointcut execsInProblemClass(): within(ProblemClass)
&amp;&amp; execution(* *(..));
</pre><p>
will pick out each execution join point of every method defined
within <tt>ProblemClass</tt>. Since advice executes
at each join point picked out by the pointcut, we can reasonably
ask which join point was reached.
</p><p>
Information about the join point that was matched is available to
advice through the special variable
<tt>thisJoinPoint</tt>, of type <a href="../api/org/aspectj/lang/JoinPoint.html" target="_top"><tt>org.aspectj.lang.JoinPoint</tt></a>.
Through this object we can access information such as</p><div class="itemizedlist"><ul compact><li><a name="d0e2074"></a>
the kind of join point that was matched
</li><li><a name="d0e2076"></a>
the source location of the code associated with the join point
</li><li><a name="d0e2078"></a>
normal, short and long string representations of the
current join point
</li><li><a name="d0e2080"></a>
the actual argument values of the join point
</li><li><a name="d0e2082"></a>
the signature of the member associated with the join point
</li><li><a name="d0e2084"></a>the currently executing object</li><li><a name="d0e2086"></a>the target object</li><li><a name="d0e2088"></a>
an object encapsulating the static information about the join
point. This is also available through the special variable
<tt>thisJoinPointStaticPart</tt>.</li></ul></div><div class="sect3"><a name="d0e2093"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2093"></a>The <tt>Demo</tt> class</h4></div></div><p>The class <tt>tjp.Demo</tt> in
<tt>tjp/Demo.java</tt> defines two methods
<tt>foo</tt> and <tt>bar</tt> with different
parameter lists and return types. Both are called, with suitable
arguments, by <tt>Demo</tt>'s
<tt>go</tt> method which was invoked from within its
<tt>main</tt> method.
</p><pre class="programlisting">
public class Demo {
static Demo d;
public static void main(String[] args){
new Demo().go();
}
void go(){
d = new Demo();
d.foo(1,d);
System.out.println(d.bar(new Integer(3)));
}
void foo(int i, Object o){
System.out.println("Demo.foo(" + i + ", " + o + ")\n");
}
String bar (Integer j){
System.out.println("Demo.bar(" + j + ")\n");
return "Demo.bar(" + j + ")";
}
}
</pre></div><div class="sect3"><a name="d0e2124"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2124"></a>The <tt>GetInfo</tt> aspect</h4></div></div><p>
This aspect uses around advice to intercept the execution of
methods <tt>foo</tt> and <tt>bar</tt> in
<tt>Demo</tt>, and prints out information garnered
from <tt>thisJoinPoint</tt> to the console.
</p><pre class="programlisting">
aspect GetInfo {
static final void println(String s){ System.out.println(s); }
pointcut goCut(): cflow(this(Demo) &amp;&amp; execution(void go()));
pointcut demoExecs(): within(Demo) &amp;&amp; execution(* *(..));
Object around(): demoExecs() &amp;&amp; !execution(* go()) &amp;&amp; goCut() {
println("Intercepted message: " +
thisJoinPointStaticPart.getSignature().getName());
println("in class: " +
thisJoinPointStaticPart.getSignature().getDeclaringType().getName());
printParameters(thisJoinPoint);
println("Running original method: \n" );
Object result = proceed();
println(" result: " + result );
return result;
}
static private void printParameters(JoinPoint jp) {
println("Arguments: " );
Object[] args = jp.getArgs();
String[] names = ((CodeSignature)jp.getSignature()).getParameterNames();
Class[] types = ((CodeSignature)jp.getSignature()).getParameterTypes();
for (int i = 0; i &lt; args.length; i++) {
println(" " + i + ". " + names[i] +
" : " + types[i].getName() +
" = " + args[i]);
}
}
}
</pre><div class="sect4"><a name="d0e2146"></a><div class="titlepage"><div><h5 class="title"><a name="d0e2146"></a>Defining the scope of a pointcut</h5></div></div><p>The pointcut <tt>goCut</tt> is defined as
<pre class="programlisting">
cflow(this(Demo)) &amp;&amp; execution(void go())
</pre>
so that only executions made in the control flow of
<tt>Demo.go</tt> are intercepted. The control flow
from the method <tt>go</tt> includes the execution of
<tt>go</tt> itself, so the definition of the around
advice includes <tt>!execution(* go())</tt> to
exclude it from the set of executions advised. </p></div><div class="sect4"><a name="d0e2169"></a><div class="titlepage"><div><h5 class="title"><a name="d0e2169"></a>Printing the class and method name</h5></div></div><p>
The name of the method and that method's defining class are
available as parts of the <a href="../api/org/aspectj/lang/Signature.html" target="_top">org.aspectj.lang.Signature</a>
object returned by calling <tt>getSignature()</tt> on
either <tt>thisJoinPoint</tt> or
<tt>thisJoinPointStaticPart</tt>.
</p></div><div class="sect4"><a name="d0e2186"></a><div class="titlepage"><div><h5 class="title"><a name="d0e2186"></a>Printing the parameters</h5></div></div><p>
The static portions of the parameter details, the name and
types of the parameters, can be accessed through the <a href="../api/org/aspectj/lang/reflect/CodeSignature.html" target="_top"><tt>org.aspectj.lang.reflect.CodeSignature</tt></a>
associated with the join point. All execution join points have code
signatures, so the cast to <tt>CodeSignature</tt>
cannot fail. </p><p>
The dynamic portions of the parameter details, the actual
values of the parameters, are accessed directly from the
execution join point object.
</p></div></div></div><div class="sect2"><a name="examples-roles"></a><div class="titlepage"><div><h3 class="title"><a name="examples-roles"></a>Roles and Views</h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/introduction</tt>.)
</p><p>
Like advice, inter-type declarations are members of an aspect. They
declare members that act as if they were defined on another class.
Unlike advice, inter-type declarations affect not only the behavior
of the application, but also the structural relationship between an
application's classes.
</p><p>
This is crucial: Publically affecting the class structure of an
application makes these modifications available to other components
of the application.
</p><p>
Aspects can declare inter-type
<div class="itemizedlist"><ul compact><li><a name="d0e2218"></a>fields</li><li><a name="d0e2220"></a>methods</li><li><a name="d0e2222"></a>constructors</li></ul></div>
and can also declare that target types
<div class="itemizedlist"><ul compact><li><a name="d0e2226"></a>implement new interfaces</li><li><a name="d0e2228"></a>extend new classes</li></ul></div>
</p><p>
This example provides three illustrations of the use of inter-type
declarations to encapsulate roles or views of a class. The class
our aspect will be dealing with, <tt>Point</tt>, is a
simple class with rectangular and polar coordinates. Our inter-type
declarations will make the class <tt>Point</tt>, in
turn, cloneable, hashable, and comparable. These facilities are
provided by AspectJ without having to modify the code for the class
<tt>Point</tt>.
</p><div class="sect3"><a name="d0e2242"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2242"></a>The <tt>Point</tt> class</h4></div></div><p>The <tt>Point</tt> class defines geometric points
whose interface includes polar and rectangular coordinates, plus some
simple operations to relocate points. <tt>Point</tt>'s
implementation has attributes for both its polar and rectangular
coordinates, plus flags to indicate which currently reflect the
position of the point. Some operations cause the polar coordinates to
be updated from the rectangular, and some have the opposite effect.
This implementation, which is in intended to give the minimum number
of conversions between coordinate systems, has the property that not
all the attributes stored in a <tt>Point</tt> object
are necessary to give a canonical representation such as might be
used for storing, comparing, cloning or making hash codes from
points. Thus the aspects, though simple, are not totally trivial.
</p><p>
The diagram below gives an overview of the aspects and their
interaction with the class <tt>Point</tt>.</p><p>
<span class="inlinemediaobject"><img src="aspects.gif"></span>
</p><p></p></div><div class="sect3"><a name="d0e2271"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2271"></a>The <tt>CloneablePoint</tt> aspect</h4></div></div><p>
This first aspect is responsible for
<tt>Point</tt>'s implementation of the
<tt>Cloneable</tt> interface. It declares that
<tt>Point implements Cloneable</tt> with a
<tt>declare parents</tt> form, and also publically
declares a specialized <tt>Point</tt>'s
<tt>clone()</tt> method. In Java, all objects inherit
the method <tt>clone</tt> from the class
<tt>Object</tt>, but an object is not cloneable
unless its class also implements the interface
<tt>Cloneable</tt>. In addition, classes
frequently have requirements over and above the simple
bit-for-bit copying that <tt>Object.clone</tt> does. In
our case, we want to update a <tt>Point</tt>'s
coordinate systems before we actually clone the
<tt>Point</tt>. So our aspect makes sure that
<tt>Point</tt> overrides
<tt>Object.clone</tt> with a new method that does what
we want.
</p><p>
We also define a test <tt>main</tt> method in the
aspect for convenience.
</p><pre class="programlisting">
public aspect CloneablePoint {
declare parents: Point implements Cloneable;
public Object Point.clone() throws CloneNotSupportedException {
// we choose to bring all fields up to date before cloning.
makeRectangular();
makePolar();
return super.clone();
}
public static void main(String[] args){
Point p1 = new Point();
Point p2 = null;
p1.setPolar(Math.PI, 1.0);
try {
p2 = (Point)p1.clone();
} catch (CloneNotSupportedException e) {}
System.out.println("p1 =" + p1 );
System.out.println("p2 =" + p2 );
p1.rotate(Math.PI / -2);
System.out.println("p1 =" + p1 );
System.out.println("p2 =" + p2 );
}
}
</pre></div><div class="sect3"><a name="d0e2328"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2328"></a>The <tt>ComparablePoint</tt> aspect</h4></div></div><p>
<tt>ComparablePoint</tt> is responsible for
<tt>Point</tt>'s implementation of the
<tt>Comparable</tt> interface. </p><p>
The interface <tt>Comparable</tt> defines the
single method <tt>compareTo</tt> which can be use to define
a natural ordering relation among the objects of a class that
implement it.
</p><p>
<tt>ComparablePoint</tt> uses <tt>declare
parents</tt> to declare that <tt>Point implements
Comparable</tt>, and also publically declares the
appropriate <tt>compareTo(Object)</tt> method: A
<tt>Point</tt> <tt>p1</tt> is said to be
less than another <tt>Point</tt><tt>
p2</tt> if <tt>p1</tt> is closer to the
origin.
</p><p>
We also define a test <tt>main</tt> method in the
aspect for convenience.
</p><pre class="programlisting">
public aspect ComparablePoint {
declare parents: Point implements Comparable;
public int Point.compareTo(Object o) {
return (int) (this.getRho() - ((Point)o).getRho());
}
public static void main(String[] args){
Point p1 = new Point();
Point p2 = new Point();
System.out.println("p1 =?= p2 :" + p1.compareTo(p2));
p1.setRectangular(2,5);
p2.setRectangular(2,5);
System.out.println("p1 =?= p2 :" + p1.compareTo(p2));
p2.setRectangular(3,6);
System.out.println("p1 =?= p2 :" + p1.compareTo(p2));
p1.setPolar(Math.PI, 4);
p2.setPolar(Math.PI, 4);
System.out.println("p1 =?= p2 :" + p1.compareTo(p2));
p1.rotate(Math.PI / 4.0);
System.out.println("p1 =?= p2 :" + p1.compareTo(p2));
p1.offset(1,1);
System.out.println("p1 =?= p2 :" + p1.compareTo(p2));
}
}
</pre></div><div class="sect3"><a name="d0e2388"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2388"></a>The <tt>HashablePoint</tt> aspect</h4></div></div><p>
Our third aspect is responsible for <tt>Point</tt>'s
overriding of <tt>Object</tt>'s
<tt>equals</tt> and <tt>hashCode</tt> methods
in order to make <tt>Point</tt>s hashable.
</p><p>
The method <tt>Object.hashCode</tt> returns an
integer, suitable for use as a hash table key. It is not required
that two objects which are not equal (according to the
<tt>equals</tt> method) return different integer
results from <tt>hashCode</tt> but it can
improve performance when the integer is used as a key into a
data structure. However, any two objects which are equal
must return the same integer value from a call to
<tt>hashCode</tt>. Since the default implementation
of <tt>Object.equals</tt> returns <tt>true</tt>
only when two objects are identical, we need to redefine both
<tt>equals</tt> and <tt>hashCode</tt> to work
correctly with objects of type <tt>Point</tt>. For
example, we want two <tt>Point</tt> objects to test
equal when they have the same <tt>x</tt> and
<tt>y</tt> values, or the same <tt>rho</tt> and
<tt>theta</tt> values, not just when they refer to the same
object. We do this by overriding the methods
<tt>equals</tt> and <tt>hashCode</tt> in the
class <tt>Point</tt>.
</p><p>
So <tt>HashablePoint</tt> declares
<tt>Point</tt>'s <tt>hashCode</tt> and
<tt>equals</tt> methods, using
<tt>Point</tt>'s rectangular coordinates to
generate a hash code and to test for equality. The
<tt>x</tt> and <tt>y</tt> coordinates are
obtained using the appropriate get methods, which ensure the
rectangular coordinates are up-to-date before returning their
values.
</p><p>
And again, we supply a <tt>main</tt> method in the
aspect for testing.
</p><pre class="programlisting">
public aspect HashablePoint {
public int Point.hashCode() {
return (int) (getX() + getY() % Integer.MAX_VALUE);
}
public boolean Point.equals(Object o) {
if (o == this) { return true; }
if (!(o instanceof Point)) { return false; }
Point other = (Point)o;
return (getX() == other.getX()) &amp;&amp; (getY() == other.getY());
}
public static void main(String[] args) {
Hashtable h = new Hashtable();
Point p1 = new Point();
p1.setRectangular(10, 10);
Point p2 = new Point();
p2.setRectangular(10, 10);
System.out.println("p1 = " + p1);
System.out.println("p2 = " + p2);
System.out.println("p1.hashCode() = " + p1.hashCode());
System.out.println("p2.hashCode() = " + p2.hashCode());
h.put(p1, "P1");
System.out.println("Got: " + h.get(p2));
}
}
</pre></div></div></div><div class="sect1"><a name="examples-development"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="examples-development"></a>Development Aspects</h2></div></div><div class="sect2"><a name="tracing-using-aspects"></a><div class="titlepage"><div><h3 class="title"><a name="tracing-using-aspects"></a>Tracing using aspects</h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/tracing</tt>.)
</p><p>
Writing a class that provides tracing functionality is easy: a
couple of functions, a boolean flag for turning tracing on and
off, a choice for an output stream, maybe some code for
formatting the output -- these are all elements that
<tt>Trace</tt> classes have been known to
have. <tt>Trace</tt> classes may be highly
sophisticated, too, if the task of tracing the execution of a
program demands it.
</p><p>
But developing the support for tracing is just one part of the
effort of inserting tracing into a program, and, most likely, not
the biggest part. The other part of the effort is calling the
tracing functions at appropriate times. In large systems, this
interaction with the tracing support can be overwhelming. Plus,
tracing is one of those things that slows the system down, so
these calls should often be pulled out of the system before the
product is shipped. For these reasons, it is not unusual for
developers to write ad-hoc scripting programs that rewrite the
source code by inserting/deleting trace calls before and after
the method bodies.
</p><p>
AspectJ can be used for some of these tracing concerns in a less
ad-hoc way. Tracing can be seen as a concern that crosscuts the
entire system and as such is amenable to encapsulation in an
aspect. In addition, it is fairly independent of what the system
is doing. Therefore tracing is one of those kind of system
aspects that can potentially be plugged in and unplugged without
any side-effects in the basic functionality of the system.
</p><div class="sect3"><a name="d0e2521"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2521"></a>An Example Application</h4></div></div><p>
Throughout this example we will use a simple application that
contains only four classes. The application is about shapes. The
<tt>TwoDShape</tt> class is the root of the shape
hierarchy:
</p><pre class="programlisting">
public abstract class TwoDShape {
protected double x, y;
protected TwoDShape(double x, double y) {
this.x = x; this.y = y;
}
public double getX() { return x; }
public double getY() { return y; }
public double distance(TwoDShape s) {
double dx = Math.abs(s.getX() - x);
double dy = Math.abs(s.getY() - y);
return Math.sqrt(dx*dx + dy*dy);
}
public abstract double perimeter();
public abstract double area();
public String toString() {
return (" @ (" + String.valueOf(x) + ", " + String.valueOf(y) + ") ");
}
}
</pre><p>
<tt>TwoDShape</tt> has two subclasses,
<tt>Circle</tt> and <tt>Square</tt>:
</p><pre class="programlisting">
public class Circle extends TwoDShape {
protected double r;
public Circle(double x, double y, double r) {
super(x, y); this.r = r;
}
public Circle(double x, double y) { this( x, y, 1.0); }
public Circle(double r) { this(0.0, 0.0, r); }
public Circle() { this(0.0, 0.0, 1.0); }
public double perimeter() {
return 2 * Math.PI * r;
}
public double area() {
return Math.PI * r*r;
}
public String toString() {
return ("Circle radius = " + String.valueOf(r) + super.toString());
}
}
</pre><pre class="programlisting">
public class Square extends TwoDShape {
protected double s; // side
public Square(double x, double y, double s) {
super(x, y); this.s = s;
}
public Square(double x, double y) { this( x, y, 1.0); }
public Square(double s) { this(0.0, 0.0, s); }
public Square() { this(0.0, 0.0, 1.0); }
public double perimeter() {
return 4 * s;
}
public double area() {
return s*s;
}
public String toString() {
return ("Square side = " + String.valueOf(s) + super.toString());
}
}
</pre><p>
To run this application, compile the classes. You can do it with or
without ajc, the AspectJ compiler. If you've installed AspectJ, go
to the directory
<tt><i><tt>InstallDir</tt></i>/examples</tt>
and type:
</p><pre class="programlisting">
ajc -argfile tracing/notrace.lst
</pre><p>To run the program, type</p><pre class="programlisting">
java tracing.ExampleMain
</pre><p>(we don't need anything special on the classpath since this is pure
Java code). You should see the following output:</p><pre class="programlisting">
c1.perimeter() = 12.566370614359172
c1.area() = 12.566370614359172
s1.perimeter() = 4.0
s1.area() = 1.0
c2.distance(c1) = 4.242640687119285
s1.distance(c1) = 2.23606797749979
s1.toString(): Square side = 1.0 @ (1.0, 2.0)
</pre></div><div class="sect3"><a name="d0e2563"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2563"></a>Tracing&#8212;Version 1</h4></div></div><p>
In a first attempt to insert tracing in this application, we will
start by writing a <tt>Trace</tt> class that is
exactly what we would write if we didn't have aspects. The
implementation is in <tt>version1/Trace.java</tt>. Its
public interface is:
</p><pre class="programlisting">
public class Trace {
public static int TRACELEVEL = 0;
public static void initStream(PrintStream s) {...}
public static void traceEntry(String str) {...}
public static void traceExit(String str) {...}
}
</pre><p>
If we didn't have AspectJ, we would have to insert calls to
<tt>traceEntry</tt> and <tt>traceExit</tt> in
all methods and constructors we wanted to trace, and to initialize
<tt>TRACELEVEL</tt> and the stream. If we wanted to trace
all the methods and constructors in our example, that would amount
to around 40 calls, and we would hope we had not forgotten any
method. But we can do that more consistently and reliably with the
following aspect (found in
<tt>version1/TraceMyClasses.java</tt>):
</p><pre class="programlisting">
aspect TraceMyClasses {
pointcut myClass(): within(TwoDShape) || within(Circle) || within(Square);
pointcut myConstructor(): myClass() &amp;&amp; execution(new(..));
pointcut myMethod(): myClass() &amp;&amp; execution(* *(..));
before (): myConstructor() {
Trace.traceEntry("" + thisJoinPointStaticPart.getSignature());
}
after(): myConstructor() {
Trace.traceExit("" + thisJoinPointStaticPart.getSignature());
}
before (): myMethod() {
Trace.traceEntry("" + thisJoinPointStaticPart.getSignature());
}
after(): myMethod() {
Trace.traceExit("" + thisJoinPointStaticPart.getSignature());
}
}</pre><p>
This aspect performs the tracing calls at appropriate
times. According to this aspect, tracing is performed at the
entrance and exit of every method and constructor defined within
the shape hierarchy.
</p><p>
What is printed at before and after each of the traced join points
is the signature of the method executing. Since the signature is
static information, we can get it through
<tt>thisJoinPointStaticPart</tt>.
</p><p>
To run this version of tracing, go to the directory
<tt><i><tt>InstallDir</tt></i>/examples</tt>
and type:
</p><pre class="programlisting">
ajc -argfile tracing/tracev1.lst
</pre><p>
Running the main method of
<tt>tracing.version1.TraceMyClasses</tt> should produce
the output:
</p><pre class="programlisting">
--&gt; tracing.TwoDShape(double, double)
&lt;-- tracing.TwoDShape(double, double)
--&gt; tracing.Circle(double, double, double)
&lt;-- tracing.Circle(double, double, double)
--&gt; tracing.TwoDShape(double, double)
&lt;-- tracing.TwoDShape(double, double)
--&gt; tracing.Circle(double, double, double)
&lt;-- tracing.Circle(double, double, double)
--&gt; tracing.Circle(double)
&lt;-- tracing.Circle(double)
--&gt; tracing.TwoDShape(double, double)
&lt;-- tracing.TwoDShape(double, double)
--&gt; tracing.Square(double, double, double)
&lt;-- tracing.Square(double, double, double)
--&gt; tracing.Square(double, double)
&lt;-- tracing.Square(double, double)
--&gt; double tracing.Circle.perimeter()
&lt;-- double tracing.Circle.perimeter()
c1.perimeter() = 12.566370614359172
--&gt; double tracing.Circle.area()
&lt;-- double tracing.Circle.area()
c1.area() = 12.566370614359172
--&gt; double tracing.Square.perimeter()
&lt;-- double tracing.Square.perimeter()
s1.perimeter() = 4.0
--&gt; double tracing.Square.area()
&lt;-- double tracing.Square.area()
s1.area() = 1.0
--&gt; double tracing.TwoDShape.distance(TwoDShape)
--&gt; double tracing.TwoDShape.getX()
&lt;-- double tracing.TwoDShape.getX()
--&gt; double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.distance(TwoDShape)
c2.distance(c1) = 4.242640687119285
--&gt; double tracing.TwoDShape.distance(TwoDShape)
--&gt; double tracing.TwoDShape.getX()
&lt;-- double tracing.TwoDShape.getX()
--&gt; double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.distance(TwoDShape)
s1.distance(c1) = 2.23606797749979
--&gt; String tracing.Square.toString()
--&gt; String tracing.TwoDShape.toString()
&lt;-- String tracing.TwoDShape.toString()
&lt;-- String tracing.Square.toString()
s1.toString(): Square side = 1.0 @ (1.0, 2.0)
</pre><p>
When <tt>TraceMyClasses.java</tt> is not provided to
<b>ajc</b>, the aspect does not have any affect on the
system and the tracing is unplugged.
</p></div><div class="sect3"><a name="d0e2623"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2623"></a>Tracing&#8212;Version 2</h4></div></div><p>
Another way to accomplish the same thing would be to write a
reusable tracing aspect that can be used not only for these
application classes, but for any class. One way to do this is to
merge the tracing functionality of
<tt>Trace&#8212;version1</tt> with the crosscutting
support of <tt>TraceMyClasses&#8212;version1</tt>. We end
up with a <tt>Trace</tt> aspect (found in
<tt>version2/Trace.java</tt>) with the following public
interface
</p><pre class="programlisting">
abstract aspect Trace {
public static int TRACELEVEL = 2;
public static void initStream(PrintStream s) {...}
protected static void traceEntry(String str) {...}
protected static void traceExit(String str) {...}
abstract pointcut myClass();
}
</pre><p>
In order to use it, we need to define our own subclass that knows
about our application classes, in
<tt>version2/TraceMyClasses.java</tt>:
</p><pre class="programlisting">
public aspect TraceMyClasses extends Trace {
pointcut myClass(): within(TwoDShape) || within(Circle) || within(Square);
public static void main(String[] args) {
Trace.TRACELEVEL = 2;
Trace.initStream(System.err);
ExampleMain.main(args);
}
}
</pre><p>
Notice that we've simply made the pointcut
<tt>classes</tt>, that was an abstract pointcut in the
super-aspect, concrete. To run this version of tracing, go to the
directory <tt>examples</tt> and type:
</p><pre class="programlisting">
ajc -argfile tracing/tracev2.lst
</pre><p>
The file tracev2.lst lists the application classes as well as this
version of the files Trace.java and TraceMyClasses.java. Running
the main method of
<tt>tracing.version2.TraceMyClasses</tt> should
output exactly the same trace information as that from version 1.
</p><p>
The entire implementation of the new <tt>Trace</tt>
class is:
</p><pre class="programlisting">
abstract aspect Trace {
// implementation part
public static int TRACELEVEL = 2;
protected static PrintStream stream = System.err;
protected static int callDepth = 0;
public static void initStream(PrintStream s) {
stream = s;
}
protected static void traceEntry(String str) {
if (TRACELEVEL == 0) return;
if (TRACELEVEL == 2) callDepth++;
printEntering(str);
}
protected static void traceExit(String str) {
if (TRACELEVEL == 0) return;
printExiting(str);
if (TRACELEVEL == 2) callDepth--;
}
private static void printEntering(String str) {
printIndent();
stream.println("--&gt; " + str);
}
private static void printExiting(String str) {
printIndent();
stream.println("&lt;-- " + str);
}
private static void printIndent() {
for (int i = 0; i &lt; callDepth; i++)
stream.print(" ");
}
// protocol part
abstract pointcut myClass();
pointcut myConstructor(): myClass() &amp;&amp; execution(new(..));
pointcut myMethod(): myClass() &amp;&amp; execution(* *(..));
before(): myConstructor() {
traceEntry("" + thisJoinPointStaticPart.getSignature());
}
after(): myConstructor() {
traceExit("" + thisJoinPointStaticPart.getSignature());
}
before(): myMethod() {
traceEntry("" + thisJoinPointStaticPart.getSignature());
}
after(): myMethod() {
traceExit("" + thisJoinPointStaticPart.getSignature());
}
}
</pre><p>
This version differs from version 1 in several subtle ways. The
first thing to notice is that this <tt>Trace</tt>
class merges the functional part of tracing with the crosscutting
of the tracing calls. That is, in version 1, there was a sharp
separation between the tracing support (the class
<tt>Trace</tt>) and the crosscutting usage of it (by
the class <tt>TraceMyClasses</tt>). In this version
those two things are merged. That's why the description of this
class explicitly says that "Trace messages are printed before and
after constructors and methods are," which is what we wanted in the
first place. That is, the placement of the calls, in this version,
is established by the aspect class itself, leaving less opportunity
for misplacing calls.</p><p>
A consequence of this is that there is no need for providing
<tt>traceEntry</tt> and <tt>traceExit</tt> as
public operations of this class. You can see that they were
classified as protected. They are supposed to be internal
implementation details of the advice.
</p><p>
The key piece of this aspect is the abstract pointcut classes that
serves as the base for the definition of the pointcuts constructors
and methods. Even though <tt>classes</tt> is
abstract, and therefore no concrete classes are mentioned, we can
put advice on it, as well as on the pointcuts that are based on
it. The idea is "we don't know exactly what the pointcut will be,
but when we do, here's what we want to do with it." In some ways,
abstract pointcuts are similar to abstract methods. Abstract
methods don't provide the implementation, but you know that the
concrete subclasses will, so you can invoke those methods.
</p></div></div></div><div class="sect1"><a name="examples-production"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="examples-production"></a>Production Aspects</h2></div></div><div class="sect2"><a name="a-bean-aspect"></a><div class="titlepage"><div><h3 class="title"><a name="a-bean-aspect"></a>A Bean Aspect</h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/bean</tt>.)
</p><p>
This example examines an aspect that makes Point objects into
Java beans with bound properties.
</p><p>
Java beans are reusable software components that can be visually
manipulated in a builder tool. The requirements for an object to be
a bean are few. Beans must define a no-argument constructor and
must be either <tt>Serializable</tt> or
<tt>Externalizable</tt>. Any properties of the object
that are to be treated as bean properties should be indicated by
the presence of appropriate <tt>get</tt> and
<tt>set</tt> methods whose names are
<tt>get</tt><span class="emphasis"><i>property</i></span> and
<tt>set </tt><span class="emphasis"><i>property</i></span> where
<span class="emphasis"><i>property</i></span> is the name of a field in the bean
class. Some bean properties, known as bound properties, fire events
whenever their values change so that any registered listeners (such
as, other beans) will be informed of those changes. Making a bound
property involves keeping a list of registered listeners, and
creating and dispatching event objects in methods that change the
property values, such as set<span class="emphasis"><i>property</i></span>
methods.
</p><p>
<tt>Point</tt> is a simple class representing points
with rectangular coordinates. <tt>Point</tt> does not
know anything about being a bean: there are set methods for
<tt>x</tt> and <tt>y</tt> but they do not fire
events, and the class is not serializable. Bound is an aspect that
makes <tt>Point</tt> a serializable class and makes
its <tt>get</tt> and <tt>set</tt> methods
support the bound property protocol.
</p><div class="sect3"><a name="d0e2767"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2767"></a>The <tt>Point</tt> class</h4></div></div><p>
The <tt>Point</tt> class is a very simple class with
trivial getters and setters, and a simple vector offset method.
</p><pre class="programlisting">
class Point {
protected int x = 0;
protected int y = 0;
public int getX() {
return x;
}
public int getY() {
return y;
}
public void setRectangular(int newX, int newY) {
setX(newX);
setY(newY);
}
public void setX(int newX) {
x = newX;
}
public void setY(int newY) {
y = newY;
}
public void offset(int deltaX, int deltaY) {
setRectangular(x + deltaX, y + deltaY);
}
public String toString() {
return "(" + getX() + ", " + getY() + ")" ;
}
}
</pre></div><div class="sect3"><a name="d0e2780"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2780"></a>The <tt>BoundPoint</tt> aspect</h4></div></div><p>
The <tt>BoundPoint</tt> aspect is responsible for
<tt>Point</tt>'s "beanness". The first thing it does is
privately declare that each <tt>Point</tt> has a
<tt>support</tt> field that holds reference to an
instance of <tt>PropertyChangeSupport</tt>.
<pre class="programlisting">
private PropertyChangeSupport Point.support = new PropertyChangeSupport(this);
</pre>
The property change support object must be constructed with a
reference to the bean for which it is providing support, so it is
initialized by passing it <tt>this</tt>, an instance of
<tt>Point</tt>. Since the <tt>support</tt>
field is private declared in the aspect, only the code in the
aspect can refer to it.
</p><p>
The aspect also declares <tt>Point</tt>'s methods for
registering and managing listeners for property change events,
which delegate the work to the property change support object:
<pre class="programlisting">
public void Point.addPropertyChangeListener(PropertyChangeListener listener){
support.addPropertyChangeListener(listener);
}
public void Point.addPropertyChangeListener(String propertyName,
PropertyChangeListener listener){
support.addPropertyChangeListener(propertyName, listener);
}
public void Point.removePropertyChangeListener(String propertyName,
PropertyChangeListener listener) {
support.removePropertyChangeListener(propertyName, listener);
}
public void Point.removePropertyChangeListener(PropertyChangeListener listener) {
support.removePropertyChangeListener(listener);
}
public void Point.hasListeners(String propertyName) {
support.hasListeners(propertyName);
}
</pre>
</p><p>
The aspect is also responsible for making sure
<tt>Point</tt> implements the
<tt>Serializable</tt> interface:
<pre class="programlisting">
declare parents: Point implements Serializable;
</pre>
Implementing this interface in Java does not require any methods to
be implemented. Serialization for <tt>Point</tt>
objects is provided by the default serialization method.
</p><p>
The <tt>setters</tt> pointcut picks out calls to the
<tt>Point</tt>'s <tt>set</tt> methods: any
method whose name begins with "<tt>set</tt>" and takes
one parameter. The around advice on <tt>setters()</tt>
stores the values of the <tt>X</tt> and
<tt>Y</tt> properties, calls the original
<tt>set</tt> method and then fires the appropriate
property change event according to which set method was
called.
</p><pre class="programlisting">
aspect BoundPoint {
private PropertyChangeSupport Point.support = new PropertyChangeSupport(this);
public void Point.addPropertyChangeListener(PropertyChangeListener listener){
support.addPropertyChangeListener(listener);
}
public void Point.addPropertyChangeListener(String propertyName,
PropertyChangeListener listener){
support.addPropertyChangeListener(propertyName, listener);
}
public void Point.removePropertyChangeListener(String propertyName,
PropertyChangeListener listener) {
support.removePropertyChangeListener(propertyName, listener);
}
public void Point.removePropertyChangeListener(PropertyChangeListener listener) {
support.removePropertyChangeListener(listener);
}
public void Point.hasListeners(String propertyName) {
support.hasListeners(propertyName);
}
declare parents: Point implements Serializable;
pointcut setter(Point p): call(void Point.set*(*)) &amp;&amp; target(p);
void around(Point p): setter(p) {
String propertyName =
thisJoinPointStaticPart.getSignature().getName().substring("set".length());
int oldX = p.getX();
int oldY = p.getY();
proceed(p);
if (propertyName.equals("X")){
firePropertyChange(p, propertyName, oldX, p.getX());
} else {
firePropertyChange(p, propertyName, oldY, p.getY());
}
}
void firePropertyChange(Point p,
String property,
double oldval,
double newval) {
p.support.firePropertyChange(property,
new Double(oldval),
new Double(newval));
}
}
</pre></div><div class="sect3"><a name="d0e2865"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2865"></a>The Test Program</h4></div></div><p>
The test program registers itself as a property change listener to
a <tt>Point</tt> object that it creates and then performs
simple manipulation of that point: calling its set methods and the
offset method. Then it serializes the point and writes it to a file
and then reads it back. The result of saving and restoring the
point is that a new point is created.
</p><pre class="programlisting">
class Demo implements PropertyChangeListener {
static final String fileName = "test.tmp";
public void propertyChange(PropertyChangeEvent e){
System.out.println("Property " + e.getPropertyName() + " changed from " +
e.getOldValue() + " to " + e.getNewValue() );
}
public static void main(String[] args){
Point p1 = new Point();
p1.addPropertyChangeListener(new Demo());
System.out.println("p1 =" + p1);
p1.setRectangular(5,2);
System.out.println("p1 =" + p1);
p1.setX( 6 );
p1.setY( 3 );
System.out.println("p1 =" + p1);
p1.offset(6,4);
System.out.println("p1 =" + p1);
save(p1, fileName);
Point p2 = (Point) restore(fileName);
System.out.println("Had: " + p1);
System.out.println("Got: " + p2);
}
...
}
</pre></div><div class="sect3"><a name="d0e2875"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2875"></a>Compiling and Running the Example</h4></div></div><p>
To compile and run this example, go to the examples directory and type:
</p><pre class="programlisting">
ajc -argfile bean/files.lst
java bean.Demo
</pre></div></div><div class="sect2"><a name="the-subject-observer-protocol"></a><div class="titlepage"><div><h3 class="title"><a name="the-subject-observer-protocol"></a>The Subject/Observer Protocol</h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/observer</tt>.)
</p><p>
This demo illustrates how the Subject/Observer design pattern can be
coded with aspects.
</p><p>
The demo consists of the following: A colored label is a
renderable object that has a color that cycles through a set of
colors, and a number that records the number of cycles it has been
through. A button is an action item that records when it is
clicked.
</p><p>
With these two kinds of objects, we can build up a Subject/Observer
relationship in which colored labels observe the clicks of buttons;
that is, where colored labels are the observers and buttons are the
subjects.
</p><p>
The demo is designed and implemented using the Subject/Observer
design pattern. The remainder of this example explains the classes
and aspects of this demo, and tells you how to run it.
</p><div class="sect3"><a name="d0e2901"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2901"></a>Generic Components</h4></div></div><p>
The generic parts of the protocol are the interfaces
<tt>Subject</tt> and <tt>Observer</tt>,
and the abstract aspect
<tt>SubjectObserverProtocol</tt>. The
<tt>Subject</tt> interface is simple, containing
methods to add, remove, and view <tt>Observer</tt>
objects, and a method for getting data about state changes:
</p><pre class="programlisting">
interface Subject {
void addObserver(Observer obs);
void removeObserver(Observer obs);
Vector getObservers();
Object getData();
}
</pre><p>
The <tt>Observer</tt> interface is just as simple,
with methods to set and get <tt>Subject</tt> objects,
and a method to call when the subject gets updated.
</p><pre class="programlisting">
interface Observer {
void setSubject(Subject s);
Subject getSubject();
void update();
}
</pre><p>
The <tt>SubjectObserverProtocol</tt> aspect contains
within it all of the generic parts of the protocol, namely, how to
fire the <tt>Observer</tt> objects' update methods
when some state changes in a subject.
</p><pre class="programlisting">
abstract aspect SubjectObserverProtocol {
abstract pointcut stateChanges(Subject s);
after(Subject s): stateChanges(s) {
for (int i = 0; i &lt; s.getObservers().size(); i++) {
((Observer)s.getObservers().elementAt(i)).update();
}
}
private Vector Subject.observers = new Vector();
public void Subject.addObserver(Observer obs) {
observers.addElement(obs);
obs.setSubject(this);
}
public void Subject.removeObserver(Observer obs) {
observers.removeElement(obs);
obs.setSubject(null);
}
public Vector Subject.getObservers() { return observers; }
private Subject Observer.subject = null;
public void Observer.setSubject(Subject s) { subject = s; }
public Subject Observer.getSubject() { return subject; }
}
</pre><p>
Note that this aspect does three things. It define an abstract
pointcut that extending aspects can override. It defines advice
that should run after the join points of the pointcut. And it
declares an inter-tpye field and two inter-type methods so that
each <tt>Observer</tt> can hold onto its <tt>Subject</tt>.
</p></div><div class="sect3"><a name="d0e2951"></a><div class="titlepage"><div><h4 class="title"><a name="d0e2951"></a>Application Classes</h4></div></div><p>
<tt>Button</tt> objects extend
<tt>java.awt.Button</tt>, and all they do is make
sure the <tt>void click()</tt> method is called whenever
a button is clicked.
</p><pre class="programlisting">
class Button extends java.awt.Button {
static final Color defaultBackgroundColor = Color.gray;
static final Color defaultForegroundColor = Color.black;
static final String defaultText = "cycle color";
Button(Display display) {
super();
setLabel(defaultText);
setBackground(defaultBackgroundColor);
setForeground(defaultForegroundColor);
addActionListener(new ActionListener() {
public void actionPerformed(ActionEvent e) {
Button.this.click();
}
});
display.addToFrame(this);
}
public void click() {}
}
</pre><p>
Note that this class knows nothing about being a Subject.
</p><p>
ColorLabel objects are labels that support the void colorCycle()
method. Again, they know nothing about being an observer.
</p><pre class="programlisting">
class ColorLabel extends Label {
ColorLabel(Display display) {
super();
display.addToFrame(this);
}
final static Color[] colors = {Color.red, Color.blue,
Color.green, Color.magenta};
private int colorIndex = 0;
private int cycleCount = 0;
void colorCycle() {
cycleCount++;
colorIndex = (colorIndex + 1) % colors.length;
setBackground(colors[colorIndex]);
setText("" + cycleCount);
}
}
</pre><p>
Finally, the <tt>SubjectObserverProtocolImpl</tt>
implements the subject/observer protocol, with
<tt>Button</tt> objects as subjects and
<tt>ColorLabel</tt> objects as observers:
</p><pre class="programlisting">
package observer;
import java.util.Vector;
aspect SubjectObserverProtocolImpl extends SubjectObserverProtocol {
declare parents: Button implements Subject;
public Object Button.getData() { return this; }
declare parents: ColorLabel implements Observer;
public void ColorLabel.update() {
colorCycle();
}
pointcut stateChanges(Subject s):
target(s) &amp;&amp;
call(void Button.click());
}</pre><p>
It does this by assuring that <tt>Button</tt> and
<tt>ColorLabel</tt> implement the appropriate
interfaces, declaring that they implement the methods required by
those interfaces, and providing a definition for the abstract
<tt>stateChanges</tt> pointcut. Now, every time a
<tt>Button</tt> is clicked, all
<tt>ColorLabel</tt> objects observing that button
will <tt>colorCycle</tt>.
</p></div><div class="sect3"><a name="d0e3006"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3006"></a>Compiling and Running</h4></div></div><p>
<tt>Demo</tt> is the top class that starts this
demo. It instantiates a two buttons and three observers and links
them together as subjects and observers. So to run the demo, go to
the <tt>examples</tt> directory and type:
</p><pre class="programlisting">
ajc -argfile observer/files.lst
java observer.Demo
</pre></div></div><div class="sect2"><a name="a-simple-telecom-simulation"></a><div class="titlepage"><div><h3 class="title"><a name="a-simple-telecom-simulation"></a>A Simple Telecom Simulation</h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/telecom</tt>.)
</p><p>
This example illustrates some ways that dependent concerns can be
encoded with aspects. It uses an example system comprising a simple
model of telephone connections to which timing and billing features
are added using aspects, where the billing feature depends upon the
timing feature.
</p><div class="sect3"><a name="d0e3032"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3032"></a>The Application</h4></div></div><p>
The example application is a simple simulation of a telephony
system in which customers make, accept, merge and hang-up both
local and long distance calls. The application architecture is in
three layers.
</p><div class="itemizedlist"><ul><li><p><a name="d0e3038"></a>
The basic objects provide basic functionality to simulate
customers, calls and connections (regular calls have one
connection, conference calls have more than one).
</p></li><li><p><a name="d0e3041"></a>
The timing feature is concerned with timing the connections
and keeping the total connection time per customer. Aspects
are used to add a timer to each connection and to manage the
total time per customer.
</p></li><li><p><a name="d0e3044"></a>
The billing feature is concerned with charging customers for
the calls they make. Aspects are used to calculate a charge
per connection and, upon termination of a connection, to add
the charge to the appropriate customer's bill. The billing
aspect builds upon the timing aspect: it uses a pointcut
defined in Timing and it uses the timers that are associated
with connections.
</p></li></ul></div><p>
The simulation of system has three configurations: basic, timing
and billing. Programs for the three configurations are in classes
<tt>BasicSimulation</tt>,
<tt>TimingSimulation</tt> and
<tt>BillingSimulation</tt>. These share a common
superclass <tt>AbstractSimulation</tt>, which
defines the method run with the simulation itself and the method
wait used to simulate elapsed time.
</p></div><div class="sect3"><a name="d0e3061"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3061"></a>The Basic Objects</h4></div></div><p>
The telecom simulation comprises the classes
<tt>Customer</tt>, <tt>Call</tt> and
the abstract class <tt>Connection</tt> with its two
concrete subclasses <tt>Local</tt> and
<tt>LongDistance</tt>. Customers have a name and a
numeric area code. They also have methods for managing
calls. Simple calls are made between one customer (the caller)
and another (the receiver), a <tt>Connection</tt>
object is used to connect them. Conference calls between more
than two customers will involve more than one connection. A
customer may be involved in many calls at one time.
<span class="inlinemediaobject"><img src="telecom.gif"></span>
</p></div><div class="sect3"><a name="d0e3088"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3088"></a>The <tt>Customer</tt> class</h4></div></div><p>
<tt>Customer</tt> has methods
<tt>call</tt>, <tt>pickup</tt>,
<tt>hangup</tt> and <tt>merge</tt> for
managing calls.
</p><pre class="programlisting">
public class Customer {
private String name;
private int areacode;
private Vector calls = new Vector();
protected void removeCall(Call c){
calls.removeElement(c);
}
protected void addCall(Call c){
calls.addElement(c);
}
public Customer(String name, int areacode) {
this.name = name;
this.areacode = areacode;
}
public String toString() {
return name + "(" + areacode + ")";
}
public int getAreacode(){
return areacode;
}
public boolean localTo(Customer other){
return areacode == other.areacode;
}
public Call call(Customer receiver) {
Call call = new Call(this, receiver);
addCall(call);
return call;
}
public void pickup(Call call) {
call.pickup();
addCall(call);
}
public void hangup(Call call) {
call.hangup(this);
removeCall(call);
}
public void merge(Call call1, Call call2){
call1.merge(call2);
removeCall(call2);
}
}
</pre></div><div class="sect3"><a name="d0e3113"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3113"></a>The <tt>Call</tt> class</h4></div></div><p>
Calls are created with a caller and receiver who are customers. If
the caller and receiver have the same area code then the call can
be established with a <tt>Local</tt> connection (see
below), otherwise a <tt>LongDistance</tt> connection
is required. A call comprises a number of connections between
customers. Initially there is only the connection between the
caller and receiver but additional connections can be added if
calls are merged to form conference calls.
</p></div><div class="sect3"><a name="d0e3127"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3127"></a>The <tt>Connection</tt> class</h4></div></div><p>
The class <tt>Connection</tt> models the physical
details of establishing a connection between customers. It does
this with a simple state machine (connections are initially
<tt>PENDING</tt>, then <tt>COMPLETED</tt> and
finally <tt>DROPPED</tt>). Messages are printed to the
console so that the state of connections can be
observed. Connection is an abstract class with two concrete
subclasses: <tt>Local</tt> and
<tt>LongDistance</tt>.
</p><pre class="programlisting">
abstract class Connection {
public static final int PENDING = 0;
public static final int COMPLETE = 1;
public static final int DROPPED = 2;
Customer caller, receiver;
private int state = PENDING;
Connection(Customer a, Customer b) {
this.caller = a;
this.receiver = b;
}
public int getState(){
return state;
}
public Customer getCaller() { return caller; }
public Customer getReceiver() { return receiver; }
void complete() {
state = COMPLETE;
System.out.println("connection completed");
}
void drop() {
state = DROPPED;
System.out.println("connection dropped");
}
public boolean connects(Customer c){
return (caller == c || receiver == c);
}
}
</pre></div><div class="sect3"><a name="d0e3155"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3155"></a>The <tt>Local</tt> and <tt>LongDistance</tt> classes</h4></div></div><p>
The two kinds of connections supported by our simulation are
<tt>Local</tt> and <tt>LongDistance</tt>
connections.
</p><pre class="programlisting">
class Local extends Connection {
Local(Customer a, Customer b) {
super(a, b);
System.out.println("[new local connection from " +
a + " to " + b + "]");
}
}
</pre><pre class="programlisting">
class LongDistance extends Connection {
LongDistance(Customer a, Customer b) {
super(a, b);
System.out.println("[new long distance connection from " +
a + " to " + b + "]");
}
}
</pre></div><div class="sect3"><a name="d0e3176"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3176"></a>Compiling and Running the Basic Simulation</h4></div></div><p>
The source files for the basic system are listed in the file
<tt>basic.lst</tt>. To build and run the basic system,
in a shell window, type these commands:
</p><pre class="programlisting">
ajc -argfile telecom/basic.lst
java telecom.BasicSimulation
</pre></div><div class="sect3"><a name="d0e3186"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3186"></a>The Timing aspect</h4></div></div><p>
The <tt>Timing</tt> aspect keeps track of total
connection time for each <tt>Customer</tt> by
starting and stopping a timer associated with each connection. It
uses some helper classes:
</p><div class="sect4"><a name="d0e3197"></a><div class="titlepage"><div><h5 class="title"><a name="d0e3197"></a>The <tt>Timer</tt> class</h5></div></div><p>
A <tt>Timer</tt> object simply records the current
time when it is started and stopped, and returns their difference
when asked for the elapsed time. The aspect
<tt>TimerLog</tt> (below) can be used to cause the
start and stop times to be printed to standard output.
</p><pre class="programlisting">
class Timer {
long startTime, stopTime;
public void start() {
startTime = System.currentTimeMillis();
stopTime = startTime;
}
public void stop() {
stopTime = System.currentTimeMillis();
}
public long getTime() {
return stopTime - startTime;
}
}
</pre></div></div><div class="sect3"><a name="d0e3213"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3213"></a>The <tt>TimerLog</tt> aspect</h4></div></div><p>
The <tt>TimerLog</tt> aspect can be included in a
build to get the timer to announce when it is started and
stopped.
</p><pre class="programlisting">
public aspect TimerLog {
after(Timer t): target(t) &amp;&amp; call(* Timer.start()) {
System.err.println("Timer started: " + t.startTime);
}
after(Timer t): target(t) &amp;&amp; call(* Timer.stop()) {
System.err.println("Timer stopped: " + t.stopTime);
}
}
</pre></div><div class="sect3"><a name="d0e3226"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3226"></a>The <tt>Timing</tt> aspect</h4></div></div><p>
The <tt>Timing</tt> aspect is declares an
inter-type field <tt>totalConnectTime</tt> for
<tt>Customer</tt> to store the accumulated connection
time per <tt>Customer</tt>. It also declares that
each <tt>Connection</tt> object has a timer.
<pre class="programlisting">
public long Customer.totalConnectTime = 0;
private Timer Connection.timer = new Timer();
</pre>
Two pieces of after advice ensure that the timer is started when
a connection is completed and and stopped when it is dropped. The
pointcut <tt>endTiming</tt> is defined so that it can
be used by the <tt>Billing</tt> aspect.
</p><pre class="programlisting">
public aspect Timing {
public long Customer.totalConnectTime = 0;
public long getTotalConnectTime(Customer cust) {
return cust.totalConnectTime;
}
private Timer Connection.timer = new Timer();
public Timer getTimer(Connection conn) { return conn.timer; }
after (Connection c): target(c) &amp;&amp; call(void Connection.complete()) {
getTimer(c).start();
}
pointcut endTiming(Connection c): target(c) &amp;&amp;
call(void Connection.drop());
after(Connection c): endTiming(c) {
getTimer(c).stop();
c.getCaller().totalConnectTime += getTimer(c).getTime();
c.getReceiver().totalConnectTime += getTimer(c).getTime();
}
}</pre></div><div class="sect3"><a name="d0e3260"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3260"></a>The <tt>Billing</tt> aspect</h4></div></div><p>
The Billing system adds billing functionality to the telecom
application on top of timing.
</p><p>
The <tt>Billing</tt> aspect declares that each
<tt>Connection</tt> has a <tt>payer</tt>
inter-type field to indicate who initiated the call and therefore
who is responsible to pay for it. It also declares the inter-type
method <tt>callRate</tt> of
<tt>Connection</tt> so that local and long distance
calls can be charged differently. The call charge must be
calculated after the timer is stopped; the after advice on pointcut
<tt>Timing.endTiming</tt> does this, and
<tt>Billing</tt> is declared to be more precedent
than <tt>Timing</tt> to make sure that this advice
runs after <tt>Timing</tt>'s advice on the same join
point. Finally, it declares inter-type methods and fields for
<tt>Customer</tt> to handle the
<tt>totalCharge</tt>.
</p><pre class="programlisting">
public aspect Billing {
// precedence required to get advice on endtiming in the right order
declare precedence: Billing, Timing;
public static final long LOCAL_RATE = 3;
public static final long LONG_DISTANCE_RATE = 10;
public Customer Connection.payer;
public Customer getPayer(Connection conn) { return conn.payer; }
after(Customer cust) returning (Connection conn):
args(cust, ..) &amp;&amp; call(Connection+.new(..)) {
conn.payer = cust;
}
public abstract long Connection.callRate();
public long LongDistance.callRate() { return LONG_DISTANCE_RATE; }
public long Local.callRate() { return LOCAL_RATE; }
after(Connection conn): Timing.endTiming(conn) {
long time = Timing.aspectOf().getTimer(conn).getTime();
long rate = conn.callRate();
long cost = rate * time;
getPayer(conn).addCharge(cost);
}
public long Customer.totalCharge = 0;
public long getTotalCharge(Customer cust) { return cust.totalCharge; }
public void Customer.addCharge(long charge){
totalCharge += charge;
}
}
</pre></div><div class="sect3"><a name="d0e3305"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3305"></a>Accessing the inter-type state</h4></div></div><p>
Both the aspects <tt>Timing</tt> and
<tt>Billing</tt> contain the definition of operations
that the rest of the system may want to access. For example, when
running the simulation with one or both aspects, we want to find
out how much time each customer spent on the telephone and how big
their bill is. That information is also stored in the classes, but
they are accessed through static methods of the aspects, since the
state they refer to is private to the aspect.
</p><p>
Take a look at the file
<tt>TimingSimulation.java</tt>. The most important
method of this class is the method
<tt>report(Customer)</tt>, which is used in the method
run of the superclass
<tt>AbstractSimulation</tt>. This method is intended
to print out the status of the customer, with respect to the
<tt>Timing</tt> feature.
</p><pre class="programlisting">
protected void report(Customer c){
Timing t = Timing.aspectOf();
System.out.println(c + " spent " + t.getTotalConnectTime(c));
}
</pre></div><div class="sect3"><a name="d0e3332"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3332"></a>Compiling and Running</h4></div></div><p>
The files timing.lst and billing.lst contain file lists for the
timing and billing configurations. To build and run the application
with only the timing feature, go to the directory examples and
type:
</p><pre class="programlisting">
ajc -argfile telecom/timing.lst
java telecom.TimingSimulation
</pre><p>
To build and run the application with the timing and billing
features, go to the directory examples and type:
</p><pre class="programlisting">
ajc -argfile telecom/billing.lst
java telecom.BillingSimulation
</pre></div><div class="sect3"><a name="d0e3343"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3343"></a>Discussion</h4></div></div><p>
There are some explicit dependencies between the aspects Billing
and Timing:
<div class="itemizedlist"><ul><li><p><a name="d0e3349"></a>
Billing is declared more precedent than Timing so that Billing's
after advice runs after that of Timing when they are on the
same join point.
</p></li><li><p><a name="d0e3352"></a>
Billing uses the pointcut Timing.endTiming.
</p></li><li><p><a name="d0e3355"></a>
Billing needs access to the timer associated with a connection.
</p></li></ul></div>
</p></div></div></div><div class="sect1"><a name="examples-reusable"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="examples-reusable"></a>Reusable Aspects</h2></div></div><div class="sect2"><a name="tracing-using-aspects-revisited"></a><div class="titlepage"><div><h3 class="title"><a name="tracing-using-aspects-revisited"></a>Tracing using Aspects, Revisited</h3></div></div><p>
(The code for this example is in
<tt><i><tt>InstallDir</tt></i>/examples/tracing</tt>.)
</p><div class="sect3"><a name="d0e3374"></a><div class="titlepage"><div><h4 class="title"><a name="d0e3374"></a>Tracing&#8212;Version 3</h4></div></div><p>
One advantage of not exposing the methods traceEntry and
traceExit as public operations is that we can easily change their
interface without any dramatic consequences in the rest of the
code.
</p><p>
Consider, again, the program without AspectJ. Suppose, for
example, that at some point later the requirements for tracing
change, stating that the trace messages should always include the
string representation of the object whose methods are being
traced. This can be achieved in at least two ways. One way is
keep the interface of the methods <tt>traceEntry</tt>
and <tt>traceExit</tt> as it was before,
</p><pre class="programlisting">
public static void traceEntry(String str);
public static void traceExit(String str);
</pre><p>
In this case, the caller is responsible for ensuring that the
string representation of the object is part of the string given
as argument. So, calls must look like:
</p><pre class="programlisting">
Trace.traceEntry("Square.distance in " + toString());
</pre><p>
Another way is to enforce the requirement with a second argument
in the trace operations, e.g.
</p><pre class="programlisting">
public static void traceEntry(String str, Object obj);
public static void traceExit(String str, Object obj);
</pre><p>
In this case, the caller is still responsible for sending the
right object, but at least there is some guarantees that some
object will be passed. The calls will look like:
</p><pre class="programlisting">
Trace.traceEntry("Square.distance", this);
</pre><p>
In either case, this change to the requirements of tracing will
have dramatic consequences in the rest of the code -- every call
to the trace operations traceEntry and traceExit must be changed!
</p><p>
Here's another advantage of doing tracing with an aspect. We've
already seen that in version 2 <tt>traceEntry</tt> and
<tt>traceExit</tt> are not publicly exposed. So
changing their interfaces, or the way they are used, has only a
small effect inside the <tt>Trace</tt>
class. Here's a partial view at the implementation of
<tt>Trace</tt>, version 3. The differences with
respect to version 2 are stressed in the comments:
</p><pre class="programlisting">
abstract aspect Trace {
public static int TRACELEVEL = 0;
protected static PrintStream stream = null;
protected static int callDepth = 0;
public static void initStream(PrintStream s) {
stream = s;
}
protected static void traceEntry(String str, Object o) {
if (TRACELEVEL == 0) return;
if (TRACELEVEL == 2) callDepth++;
printEntering(str + ": " + o.toString());
}
protected static void traceExit(String str, Object o) {
if (TRACELEVEL == 0) return;
printExiting(str + ": " + o.toString());
if (TRACELEVEL == 2) callDepth--;
}
private static void printEntering(String str) {
printIndent();
stream.println("Entering " + str);
}
private static void printExiting(String str) {
printIndent();
stream.println("Exiting " + str);
}
private static void printIndent() {
for (int i = 0; i &lt; callDepth; i++)
stream.print(" ");
}
abstract pointcut myClass(Object obj);
pointcut myConstructor(Object obj): myClass(obj) &amp;&amp; execution(new(..));
pointcut myMethod(Object obj): myClass(obj) &amp;&amp;
execution(* *(..)) &amp;&amp; !execution(String toString());
before(Object obj): myConstructor(obj) {
traceEntry("" + thisJoinPointStaticPart.getSignature(), obj);
}
after(Object obj): myConstructor(obj) {
traceExit("" + thisJoinPointStaticPart.getSignature(), obj);
}
before(Object obj): myMethod(obj) {
traceEntry("" + thisJoinPointStaticPart.getSignature(), obj);
}
after(Object obj): myMethod(obj) {
traceExit("" + thisJoinPointStaticPart.getSignature(), obj);
}
}
</pre><p>
As you can see, we decided to apply the first design by preserving
the interface of the methods <tt>traceEntry</tt> and
<tt>traceExit</tt>. But it doesn't matter&#8212;we could
as easily have applied the second design (the code in the directory
<tt>examples/tracing/version3</tt> has the second
design). The point is that the effects of this change in the
tracing requirements are limited to the
<tt>Trace</tt> aspect class.
</p><p>
One implementation change worth noticing is the specification of
the pointcuts. They now expose the object. To maintain full
consistency with the behavior of version 2, we should have included
tracing for static methods, by defining another pointcut for static
methods and advising it. We leave that as an exercise.
</p><p>
Moreover, we had to exclude the execution join point of the method
<tt>toString</tt> from the <tt>methods</tt>
pointcut. The problem here is that <tt>toString</tt> is
being called from inside the advice. Therefore if we trace it, we
will end up in an infinite recursion of calls. This is a subtle
point, and one that you must be aware when writing advice. If the
advice calls back to the objects, there is always the possibility
of recursion. Keep that in mind!
</p><p>
In fact, esimply excluding the execution join point may not be
enough, if there are calls to other traced methods within it -- in
which case, the restriction should be
</p><pre class="programlisting">
&amp;&amp; !cflow(execution(String toString()))
</pre><p>
excluding both the execution of toString methods and all join
points under that execution.
</p><p>
In summary, to implement the change in the tracing requirements we
had to make a couple of changes in the implementation of the
<tt>Trace</tt> aspect class, including changing the
specification of the pointcuts. That's only natural. But the
implementation changes were limited to this aspect. Without
aspects, we would have to change the implementation of every
application class.
</p><p>
Finally, to run this version of tracing, go to the directory
<tt>examples</tt> and type:
</p><pre class="programlisting">
ajc -argfile tracing/tracev3.lst
</pre><p>
The file tracev3.lst lists the application classes as well as this
version of the files <tt>Trace.java</tt> and
<tt>TraceMyClasses.java</tt>. To run the program, type
</p><pre class="programlisting">
java tracing.version3.TraceMyClasses
</pre><p>The output should be:</p><pre class="programlisting">
--&gt; tracing.TwoDShape(double, double)
&lt;-- tracing.TwoDShape(double, double)
--&gt; tracing.Circle(double, double, double)
&lt;-- tracing.Circle(double, double, double)
--&gt; tracing.TwoDShape(double, double)
&lt;-- tracing.TwoDShape(double, double)
--&gt; tracing.Circle(double, double, double)
&lt;-- tracing.Circle(double, double, double)
--&gt; tracing.Circle(double)
&lt;-- tracing.Circle(double)
--&gt; tracing.TwoDShape(double, double)
&lt;-- tracing.TwoDShape(double, double)
--&gt; tracing.Square(double, double, double)
&lt;-- tracing.Square(double, double, double)
--&gt; tracing.Square(double, double)
&lt;-- tracing.Square(double, double)
--&gt; double tracing.Circle.perimeter()
&lt;-- double tracing.Circle.perimeter()
c1.perimeter() = 12.566370614359172
--&gt; double tracing.Circle.area()
&lt;-- double tracing.Circle.area()
c1.area() = 12.566370614359172
--&gt; double tracing.Square.perimeter()
&lt;-- double tracing.Square.perimeter()
s1.perimeter() = 4.0
--&gt; double tracing.Square.area()
&lt;-- double tracing.Square.area()
s1.area() = 1.0
--&gt; double tracing.TwoDShape.distance(TwoDShape)
--&gt; double tracing.TwoDShape.getX()
&lt;-- double tracing.TwoDShape.getX()
--&gt; double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.distance(TwoDShape)
c2.distance(c1) = 4.242640687119285
--&gt; double tracing.TwoDShape.distance(TwoDShape)
--&gt; double tracing.TwoDShape.getX()
&lt;-- double tracing.TwoDShape.getX()
--&gt; double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.getY()
&lt;-- double tracing.TwoDShape.distance(TwoDShape)
s1.distance(c1) = 2.23606797749979
--&gt; String tracing.Square.toString()
--&gt; String tracing.TwoDShape.toString()
&lt;-- String tracing.TwoDShape.toString()
&lt;-- String tracing.Square.toString()
s1.toString(): Square side = 1.0 @ (1.0, 2.0)
</pre></div></div></div></div><div class="chapter"><div class="titlepage"><div><h2 class="title"><a name="idioms"></a>Chapter 4. Idioms</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#idioms-intro">Introduction</a></dt></dl></div><div class="sect1"><a name="idioms-intro"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="idioms-intro"></a>Introduction</h2></div></div><p>
This chapter consists of very short snippets of AspectJ code,
typically pointcuts, that are particularly evocative or useful.
This section is a work in progress.
</p><p>
Here's an example of how to enfore a rule that code in the
java.sql package can only be used from one particular package in
your system. This doesn't require any access to code in the
java.sql package.
</p><pre class="programlisting">
/* Any call to methods or constructors in java.sql */
pointcut restrictedCall():
call(* java.sql.*.*(..)) || call(java.sql.*.new(..));
/* Any code in my system not in the sqlAccess package */
pointcut illegalSource():
within(com.foo..*) &amp;&amp; !within(com.foo.sqlAccess.*);
declare error: restrictedCall() &amp;&amp; illegalSource():
"java.sql package can only be accessed from com.foo.sqlAccess";
</pre><p>Any call to an instance of a subtype of AbstractFacade whose class is
not exactly equal to AbstractFacade:</p><pre class="programlisting">
pointcut nonAbstract(AbstractFacade af):
call(* *(..))
&amp;&amp; target(af)
&amp;&amp; !if(af.getClass() == AbstractFacade.class);
</pre><p> If AbstractFacade is an abstract class or an interface, then every
instance must be of a subtype and you can replace this with: </p><pre class="programlisting">
pointcut nonAbstract(AbstractFacade af):
call(* *(..))
&amp;&amp; target(af);
</pre><p> Any call to a method which is defined by a subtype of
AbstractFacade, but which isn't defined by the type AbstractFacade itself:
</p><pre class="programlisting">
pointcut callToUndefinedMethod():
call(* AbstractFacade+.*(..))
&amp;&amp; !call(* AbstractFacade.*(..));
</pre><p> The execution of a method that is defined in the source code for a
type that is a subtype of AbstractFacade but not in AbstractFacade itself:
</p><pre class="programlisting">
pointcut executionOfUndefinedMethod():
execution(* *(..))
&amp;&amp; within(AbstractFacade+)
&amp;&amp; !within(AbstractFacade)
</pre></div></div><div class="chapter"><div class="titlepage"><div><h2 class="title"><a name="pitfalls"></a>Chapter 5. Pitfalls</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#pitfalls-intro">Introduction</a></dt><dt><a href="#pitfalls-infiniteLoops">Infinite loops</a></dt></dl></div><div class="sect1"><a name="pitfalls-intro"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="pitfalls-intro"></a>Introduction</h2></div></div><p>
This chapter consists of a few AspectJ programs that may lead to
surprising behavior and how to understand them.
</p></div><div class="sect1"><a name="pitfalls-infiniteLoops"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="pitfalls-infiniteLoops"></a>Infinite loops</h2></div></div><p>
Here is a Java program with peculiar behavior
</p><pre class="programlisting">
public class Main {
public static void main(String[] args) {
foo();
System.out.println("done with call to foo");
}
static void foo() {
try {
foo();
} finally {
foo();
}
}
}
</pre><p>
This program will never reach the println call, but when it aborts
may have no stack trace.
</p><p>
This silence is caused by multiple StackOverflowExceptions. First
the infinite loop in the body of the method generates one, which the
finally clause tries to handle. But this finally clause also
generates an infinite loop which the current JVMs can't handle
gracefully leading to the completely silent abort.
</p><p>
The following short aspect will also generate this behavior:
</p><pre class="programlisting">
aspect A {
before(): call(* *(..)) { System.out.println("before"); }
after(): call(* *(..)) { System.out.println("after"); }
}
</pre><p>
Why? Because the call to println is also a call matched by the
pointcut <tt>call (* *(..))</tt>. We get no output because
we used simple after() advice. If the aspect were changed to
</p><pre class="programlisting">
aspect A {
before(): call(* *(..)) { System.out.println("before"); }
after() returning: call(* *(..)) { System.out.println("after"); }
}
</pre><p>
Then at least a StackOverflowException with a stack trace would be
seen. In both cases, though, the overall problem is advice applying
within its own body.
</p><p>
There's a simple idiom to use if you ever have a worry that your
advice might apply in this way. Just restrict the advice from occurring in
join points caused within the aspect. So:
</p><pre class="programlisting">
aspect A {
before(): call(* *(..)) &amp;&amp; !within(A) { System.out.println("before"); }
after() returning: call(* *(..)) &amp;&amp; !within(A) { System.out.println("after"); }
}
</pre><p>
Other solutions might be to more closely restrict the pointcut in
other ways, for example:
</p><pre class="programlisting">
aspect A {
before(): call(* MyObject.*(..)) { System.out.println("before"); }
after() returning: call(* MyObject.*(..)) { System.out.println("after"); }
}
</pre><p>
The moral of the story is that unrestricted generic pointcuts can
pick out more join points than intended.
</p></div></div><div class="appendix"><div class="titlepage"><div><h2 class="title"><a name="quick"></a>Appendix A. AspectJ Quick Reference</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#quick-pointcuts">Pointcuts</a></dt><dt><a href="#quick-typePatterns">Type Patterns</a></dt><dt><a href="#quick-advice">Advice</a></dt><dt><a href="#quick-interType">Inter-type member declarations</a></dt><dt><a href="#quick-other">Other declarations</a></dt><dt><a href="#quick-aspectAssociations">Aspects</a></dt></dl></div><div class="sect1"><a name="quick-pointcuts"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="quick-pointcuts"></a>Pointcuts</h2></div></div><div class="informaltable" id="d0e3554"><a name="d0e3554"></a><table border="0"><colgroup><col align="left"><col align="left"></colgroup><tbody valign="top"><tr><td colspan="2"><span class="bold"><b>Methods and Constructors</b></span></td></tr><tr><td><tt>call(<i><tt>Signature</tt></i>)</tt></td><td>
every call to any method or constructor matching
<i><tt>Signature</tt></i> at the call site
</td></tr><tr><td><tt>execution(<i><tt>Signature</tt></i>)</tt></td><td>
every execution of any method or constructor matching
<i><tt>Signature</tt></i></td></tr><tr><td colspan="2"><span class="bold"><b>Fields</b></span></td></tr><tr><td><tt>get(<i><tt>Signature</tt></i>)</tt></td><td>
every reference to any field matching <i><tt>Signature</tt></i></td></tr><tr><td><tt>set(<i><tt>Signature</tt></i>)</tt></td><td>
every assignment to any field matching
<i><tt>Signature</tt></i>. The assigned value can
be exposed with an <tt>args</tt> pointcut
</td></tr><tr><td colspan="2"><span class="bold"><b>Exception Handlers</b></span></td></tr><tr><td><tt>handler(<i><tt>TypePattern</tt></i>)</tt></td><td>
every exception handler for any <tt>Throwable</tt>
type in <i><tt>TypePattern</tt></i>. The exception
value can be exposed with an <tt>args</tt> pointcut
</td></tr><tr><td colspan="2"><span class="bold"><b>Advice</b></span></td></tr><tr><td><tt>adviceexecution()</tt></td><td>
every execution of any piece of advice
</td></tr><tr><td colspan="2"><span class="bold"><b>Initialization</b></span></td></tr><tr><td><tt>staticinitialization(<i><tt>TypePattern</tt></i>)</tt></td><td>
every execution of a static initializer for any type in
<i><tt>TypePattern</tt></i></td></tr><tr><td><tt>initialization(<i><tt>Signature</tt></i>)</tt></td><td>
every initialization of an object when the first constructor
called in the type matches
<i><tt>Signature</tt></i>, encompassing the return
from the super constructor call to the return of the
first-called constructor
</td></tr><tr><td><tt>preinitialization(<i><tt>Signature</tt></i>)</tt></td><td>
every pre-initialization of an object when the first
constructor called in the type matches
<i><tt>Signature</tt></i>, encompassing the entry
of the first-called constructor to the call to the super
constructor
</td></tr><tr><td colspan="2"><span class="bold"><b>Lexical</b></span></td></tr><tr><td><tt>within(<i><tt>TypePattern</tt></i>)</tt></td><td>
every join point from code defined in a type in
<i><tt>TypePattern</tt></i></td></tr><tr><td><tt>withincode(<i><tt>Signature</tt></i>)</tt></td><td>
every join point from code defined in a method or constructor
matching <i><tt>Signature</tt></i></td></tr></tbody></table><table border="0"><colgroup><col align="left"><col align="left"></colgroup><tbody valign="top"><tr><td colspan="2"><span class="bold"><b>Instanceof checks and context exposure</b></span></td></tr><tr><td><tt>this(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>)</tt></td><td>
every join point when the currently executing object is an
instance of <i><tt>Type</tt></i> or
<i><tt>Id</tt></i>'s type
</td></tr><tr><td><tt>target(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>)</tt></td><td>
every join point when the target executing object is an
instance of <i><tt>Type</tt></i> or
<i><tt>Id</tt></i>'s type
</td></tr><tr><td><tt>args(<i><tt>Type</tt></i> or
<i><tt>Id</tt></i>, ...)</tt></td><td>
every join point when the arguments are instances of
<i><tt>Type</tt></i>s or the types of the
<i><tt>Id</tt></i>s
</td></tr><tr><td colspan="2"><span class="bold"><b>Control Flow</b></span></td></tr><tr><td><tt>cflow(<i><tt>Pointcut</tt></i>)</tt></td><td>
every join point in the control flow of each join point
<i><tt>P</tt></i> picked out by
<i><tt>Pointcut</tt></i>, including
<i><tt>P</tt></i> itself
</td></tr><tr><td><tt>cflowbelow(<i><tt>Pointcut</tt></i>)</tt></td><td>
every join point below the control flow of each join point
<i><tt>P</tt></i> picked out by
<i><tt>Pointcut</tt></i>; does not include
<i><tt>P</tt></i> itself
</td></tr><tr><td colspan="2"><span class="bold"><b>Conditional</b></span></td></tr><tr><td><tt>if(<i><tt>Expression</tt></i>)</tt></td><td>
every join point when the boolean
<i><tt>Expression</tt></i> is
<tt>true</tt></td></tr></tbody></table><table border="0"><colgroup><col align="left"><col align="left"></colgroup><tbody valign="top"><tr><td colspan="2"><span class="bold"><b>Combination</b></span></td></tr><tr><td><tt>! <i><tt>Pointcut</tt></i></tt></td><td>
every join point not picked out by
<i><tt>Pointcut</tt></i></td></tr><tr><td><tt><i><tt>Pointcut0</tt></i> &amp;&amp; <i><tt>Pointcut1</tt></i></tt></td><td>
each join point picked out by both
<i><tt>Pointcut0</tt></i> and
<i><tt>Pointcut1</tt></i></td></tr><tr><td><tt><i><tt>Pointcut0</tt></i> || <i><tt>Pointcut1</tt></i></tt></td><td>
each join point picked out by either
<i><tt>Pointcut0</tt></i> or
<i><tt>Pointcut1</tt></i></td></tr><tr><td><tt>( <i><tt>Pointcut</tt></i> )</tt></td><td>
each join point picked out by
<i><tt>Pointcut</tt></i></td></tr></tbody></table></div></div><div class="sect1"><a name="quick-typePatterns"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="quick-typePatterns"></a>Type Patterns</h2></div></div><p>
A type pattern is one of
</p><div class="informaltable" id="d0e3905"><a name="d0e3905"></a><table border="0"><colgroup><col><col></colgroup><tbody valign="top"><tr><td><i><tt>TypeNamePattern</tt></i></td><td>all types in <i><tt>TypeNamePattern</tt></i></td></tr><tr><td><i><tt>SubtypePattern</tt></i></td><td>all types in <i><tt>SubtypePattern</tt></i>, a
pattern with a +. </td></tr><tr><td><i><tt>ArrayTypePattern</tt></i></td><td>all types in <i><tt>ArrayTypePattern</tt></i>,
a pattern with one or more []s. </td></tr><tr><td><tt>!<i><tt>TypePattern</tt></i></tt></td><td>all types not in <i><tt>TypePattern</tt></i></td></tr><tr><td><tt><i><tt>TypePattern0</tt></i>
&amp;&amp; <i><tt>TypePattern1</tt></i></tt></td><td>all types in both
<i><tt>TypePattern0</tt></i> and <i><tt>TypePattern1</tt></i></td></tr><tr><td><tt><i><tt>TypePattern0</tt></i> || <i><tt>TypePattern1</tt></i></tt></td><td>all types in either
<i><tt>TypePattern0</tt></i> or <i><tt>TypePattern1</tt></i></td></tr><tr><td><tt>( <i><tt>TypePattern</tt></i> )</tt></td><td>all types in <i><tt>TypePattern</tt></i></td></tr></tbody></table></div><p>
where <i><tt>TypeNamePattern</tt></i> can either be a
plain type name, the wildcard <tt>*</tt> (indicating all
types), or an identifier with embedded <tt>*</tt> and
<tt>..</tt> wildcards.
</p><p>
An embedded <tt>*</tt> in an identifier matches any
sequence of characters, but does not match the package (or
inner-type) separator ".".
</p><p>
An embedded <tt>..</tt> in an identifier matches any
sequence of characters that starts and ends with the package (or
inner-type) separator ".".
</p></div><div class="sect1"><a name="quick-advice"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="quick-advice"></a>Advice</h2></div></div><p>
Each piece of advice is of the form
<blockquote class="blockquote"><tt>[ strictfp ] <i><tt>AdviceSpec</tt></i>
[ throws <i><tt>TypeList</tt></i> ] :
<i><tt>Pointcut</tt></i> {
<i><tt>Body</tt></i> } </tt></blockquote>
where <i><tt>AdviceSpec</tt></i> is one of
</p><div class="variablelist"><dl><dt><a name="d0e4035"></a><span class="term">
<tt>before( <i><tt>Formals</tt></i> ) </tt>
</span></dt><dd>
runs before each join point
</dd><dt><a name="d0e4046"></a><span class="term">
<tt>after( <i><tt>Formals</tt></i> ) returning
[ ( <i><tt>Formal</tt></i> ) ] </tt>
</span></dt><dd>
runs after each join point that returns normally. The
optional formal gives access to the returned value
</dd><dt><a name="d0e4060"></a><span class="term">
<tt>after( <i><tt>Formals</tt></i> ) throwing [
( <i><tt>Formal</tt></i> ) ] </tt>
</span></dt><dd>
runs after each join point that throws a
<tt>Throwable</tt>. If the optional formal is
present, runs only after each join point that throws a
<tt>Throwable</tt> of the type of
<i><tt>Formal</tt></i>, and
<i><tt>Formal</tt></i> gives access to the
<tt>Throwable</tt> exception value
</dd><dt><a name="d0e4089"></a><span class="term">
<tt>after( <i><tt>Formals</tt></i> ) </tt>
</span></dt><dd>
runs after each join point regardless of whether it returns
normally or throws a <tt>Throwable</tt></dd><dt><a name="d0e4102"></a><span class="term">
<tt><i><tt>Type</tt></i>
around( <i><tt>Formals</tt></i> ) </tt>
</span></dt><dd>
runs in place of each join point. The join point can be
executed by calling <tt>proceed</tt>, which takes
the same number and types of arguments as the around advice.
</dd></dl></div><p>
Three special variables are available inside of advice bodies:
</p><div class="variablelist"><dl><dt><a name="d0e4121"></a><span class="term">
<tt>thisJoinPoint</tt>
</span></dt><dd>
an object of type <a href="../api/org/aspectj/lang/JoinPoint.html" target="_top"><tt>org.aspectj.lang.JoinPoint</tt></a>
representing the join point at which the advice is executing.
</dd><dt><a name="d0e4133"></a><span class="term">
<tt>thisJoinPointStaticPart</tt>
</span></dt><dd>
equivalent to <tt>thisJoinPoint.getStaticPart()</tt>,
but may use fewer runtime resources.
</dd><dt><a name="d0e4144"></a><span class="term">
<tt>thisEnclosingJoinPointStaticPart</tt>
</span></dt><dd>
the static part of the dynamically enclosing join point.
</dd></dl></div></div><div class="sect1"><a name="quick-interType"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="quick-interType"></a>Inter-type member declarations</h2></div></div><p>
Each inter-type member is one of
</p><div class="variablelist"><dl><dt><a name="d0e4159"></a><span class="term">
<tt>
<i><tt>Modifiers ReturnType OnType . Id</tt></i>
( <i><tt>Formals</tt></i> )
[ throws <i><tt>TypeList</tt></i> ]
{ <i><tt>Body</tt></i> }
</tt>
</span></dt><dd>
a method on <i><tt>OnType</tt></i>.
</dd><dt><a name="d0e4182"></a><span class="term">
<tt>
abstract <i><tt>Modifiers ReturnType OnType . Id</tt></i>
( <i><tt>Formals</tt></i> )
[ throws <i><tt>TypeList</tt></i> ] ;
</tt>
</span></dt><dd>
an abstract method on <i><tt>OnType</tt></i>.
</dd><dt><a name="d0e4202"></a><span class="term">
<tt>
<i><tt>Modifiers OnType . </tt></i> new
( <i><tt>Formals</tt></i> )
[ throws <i><tt>TypeList</tt></i> ]
{ <i><tt>Body</tt></i> }
</tt>
</span></dt><dd>
a constructor on <i><tt>OnType</tt></i>.
</dd><dt><a name="d0e4225"></a><span class="term">
<tt>
<i><tt>Modifiers Type OnType . Id </tt></i>
[ = <i><tt>Expression</tt></i> ] ;
</tt>
</span></dt><dd>
a field on <i><tt>OnType</tt></i>.
</dd></dl></div></div><div class="sect1"><a name="quick-other"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="quick-other"></a>Other declarations</h2></div></div><div class="variablelist"><dl><dt><a name="d0e4247"></a><span class="term">
<tt>
declare parents :
<i><tt>TypePattern</tt></i> extends
<i><tt>Type</tt></i> ;
</tt>
</span></dt><dd>
the types in <i><tt>TypePattern</tt></i> extend
<i><tt>Type</tt></i>.
</dd><dt><a name="d0e4267"></a><span class="term">
<tt>
declare parents : <i><tt>TypePattern</tt></i>
implements <i><tt>TypeList</tt></i> ;
</tt>
</span></dt><dd>
the types in <i><tt>TypePattern</tt></i>
implement the types in <i><tt>TypeList</tt></i>.
</dd><dt><a name="d0e4287"></a><span class="term">
<tt>
declare warning : <i><tt>Pointcut</tt></i> :
<i><tt>String</tt></i> ;
</tt>
</span></dt><dd>
if any of the join points in <i><tt>Pointcut</tt></i>
possibly exist in the program, the compiler emits the warning
<i><tt>String</tt></i>.
</dd><dt><a name="d0e4307"></a><span class="term">
<tt>
declare error : <i><tt>Pointcut</tt></i> :
<i><tt>String</tt></i> ;
</tt>
</span></dt><dd>
if any of the join points in <i><tt>Pointcut</tt></i>
could possibly exist in the program, the compiler emits the
error <i><tt>String</tt></i>.
</dd><dt><a name="d0e4327"></a><span class="term">
<tt>
declare soft :
<i><tt>Type</tt></i> :
<i><tt>Pointcut</tt></i> ;
</tt>
</span></dt><dd>
any <i><tt>Type</tt></i> exception
that gets thrown at any join point picked out by
<i><tt>Pointcut</tt></i> is wrapped in <a href="../api/org/aspectj/lang/SoftException.html" target="_top"><tt>org.aspectj.lang.SoftException</tt></a>.
</dd><dt><a name="d0e4351"></a><span class="term">
<tt>
declare precedence :
<i><tt>TypePatternList</tt></i> ;
</tt>
</span></dt><dd>
at any join point where multiple pieces of advice
apply, the advice precedence at that join point is in
<i><tt>TypePatternList</tt></i> order.
</dd></dl></div></div><div class="sect1"><a name="quick-aspectAssociations"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="quick-aspectAssociations"></a>Aspects</h2></div></div><p>
Each aspect is of the form
<blockquote class="blockquote"><tt>
[ privileged ] <i><tt>Modifiers</tt></i>
aspect <i><tt>Id</tt></i>
[ extends <i><tt>Type</tt></i> ]
[ implements <i><tt>TypeList</tt></i> ]
[ <i><tt>PerClause</tt></i> ]
{ <i><tt>Body</tt></i> }
</tt></blockquote>
where <i><tt>PerClause</tt></i> defines how the aspect is
instantiated and associated (<tt>issingleton()</tt> by
default):
</p><div class="informaltable" id="d0e4399"><a name="d0e4399"></a><table border="0"><colgroup><col><col><col></colgroup><thead><tr><th align="left">PerClause</th><th align="left">Description</th><th align="left">Accessor</th></tr></thead><tbody valign="top"><tr><td>
[ <tt>issingleton()</tt> ]
</td><td>
One instance of the aspect is made. This is
the default.
</td><td><tt>aspectOf()</tt> at all join points
</td></tr><tr><td><tt>perthis(<i><tt>Pointcut</tt></i>)</tt></td><td>
An instance is associated with each object that is the
currently executing object at any join point in
<i><tt>Pointcut</tt></i>.
</td><td><tt>aspectOf(Object)</tt> at all join points
</td></tr><tr><td><tt>pertarget(<i><tt>Pointcut</tt></i>)</tt></td><td>
An instance is associated with each object that is the
target object at any join point in
<i><tt>Pointcut</tt></i>.
</td><td><tt>aspectOf(Object)</tt> at all join points
</td></tr><tr><td><tt>percflow(<i><tt>Pointcut</tt></i>)</tt></td><td>
The aspect is defined for each entrance to the control flow of
the join points defined by <i><tt>Pointcut</tt></i>. </td><td><tt>aspectOf()</tt> at join points in
<tt>cflow(<i><tt>Pointcut</tt></i>)</tt></td></tr><tr><td><tt>percflowbelow(<i><tt>Pointcut</tt></i>)</tt></td><td>
The aspect is defined for each entrance to the control flow
below the join points defined by <i><tt>Pointcut</tt></i>.
</td><td><tt>aspectOf()</tt> at join points in
<tt>cflowbelow(<i><tt>Pointcut</tt></i>)</tt></td></tr></tbody></table></div></div></div><div class="appendix"><div class="titlepage"><div><h2 class="title"><a name="semantics"></a>Appendix B. Language Semantics</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#semantics-intro">Introduction</a></dt><dt><a href="#semantics-joinPoints">Join Points</a></dt><dt><a href="#semantics-pointcuts">Pointcuts</a></dt><dd><dl><dt><a href="#pointcut-definition">Pointcut definition</a></dt><dt><a href="#context-exposure">Context exposure</a></dt><dt><a href="#primitive-pointcuts">Primitive pointcuts</a></dt><dt><a href="#signatures">Signatures</a></dt><dt><a href="#matching">Matching</a></dt><dt><a href="#type-patterns">Type patterns</a></dt><dt><a href="#pattern-summary">Pattern Summary</a></dt></dl></dd><dt><a href="#semantics-advice">Advice</a></dt><dd><dl><dt><a href="#advice-modifiers">Advice modifiers</a></dt><dt><a href="#advice-and-checked-exceptions">Advice and checked exceptions</a></dt><dt><a href="#advice-precedence">Advice precedence</a></dt><dt><a href="#reflective-access-to-the-join-point">Reflective access to the join point</a></dt></dl></dd><dt><a href="#semantics-declare">Static crosscutting</a></dt><dd><dl><dt><a href="#inter-type-member-declarations">Inter-type member declarations</a></dt><dt><a href="#access-modifiers">Access modifiers</a></dt><dt><a href="#conflicts">Conflicts</a></dt><dt><a href="#extension-and-implementation">Extension and Implementation</a></dt><dt><a href="#interfaces-with-members">Interfaces with members</a></dt><dt><a href="#warnings-and-errors">Warnings and Errors</a></dt><dt><a href="#softened-exceptions">Softened exceptions</a></dt><dt><a href="#advice-precedence">Advice Precedence</a></dt><dt><a href="#statically-determinable-pointcuts">Statically determinable pointcuts</a></dt></dl></dd><dt><a href="#semantics-aspects">Aspects</a></dt><dd><dl><dt><a href="#aspect-declaration">Aspect Declaration</a></dt><dt><a href="#aspect-extension">Aspect Extension</a></dt><dt><a href="#aspect-instantiation">Aspect instantiation</a></dt><dt><a href="#aspect-privilege">Aspect privilege</a></dt></dl></dd></dl></div><div class="sect1"><a name="semantics-intro"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="semantics-intro"></a>Introduction</h2></div></div><p>
AspectJ extends Java by overlaying a concept of join points onto the
existing Java semantics and adding a few new program elements to Java:
</p><p>
A join point is a well-defined point in the execution of a
program. These include method and constructor calls, field accesses and
others described below.
</p><p>
A pointcut picks out join points, and exposes some of the values in the
execution context of those join points. There are several primitive
pointcut designators, and others can be named and defined by the
<tt>pointcut</tt> declaration.
</p><p>
A piece of advice is code that executes at each join point in a
pointcut. Advice has access to the values exposed by the
pointcut. Advice is defined by <tt>before</tt>,
<tt>after</tt>, and <tt>around</tt> declarations.
</p><p>
Inter-type declarations form AspectJ's static crosscutting features,
that is, is code that may change the type structure of a program, by
adding to or extending interfaces and classes with new fields,
constructors, or methods. Some inter-type declarations are defined
through an extension of usual method, field, and constructor
declarations, and other declarations are made with a new
<tt>declare</tt> keyword.
</p><p>
An aspect is a crosscutting type that encapsulates pointcuts, advice,
and static crosscutting features. By type, we mean Java's notion: a
modular unit of code, with a well-defined interface, about which it is
possible to do reasoning at compile time. Aspects are defined by the
<tt>aspect</tt> declaration.
</p></div><div class="sect1"><a name="semantics-joinPoints"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="semantics-joinPoints"></a>Join Points</h2></div></div><p>
While aspects define types that crosscut, the AspectJ system does not
allow completely arbitrary crosscutting. Rather, aspects define types
that cut across principled points in a program's execution. These
principled points are called join points.
</p><p>
A join point is a well-defined point in the execution of a
program. The join points defined by AspectJ are:
</p><div class="variablelist"><dl><dt><a name="d0e4541"></a><span class="term">Method call</span></dt><dd>
When a method is called, not including super calls of
non-static methods.
</dd><dt><a name="d0e4546"></a><span class="term">Method execution</span></dt><dd>
When the body of code for an actual method executes.
</dd><dt><a name="d0e4551"></a><span class="term">Constructor call</span></dt><dd>
When an object is built and that object's initial constructor is
called (i.e., not for "super" or "this" constructor calls). The
object being constructed is returned at a constructor call join
point, so its return type is considered to be the type of the
object, and the object itself may be accessed with <tt>after
returning</tt> advice.
</dd><dt><a name="d0e4559"></a><span class="term">Constructor execution</span></dt><dd>
When the body of code for an actual constructor executes, after
its this or super constructor call. The object being constructed
is the currently executing object, and so may be accessed with
the <tt>this</tt> pointcut. The constructor execution
join point for a constructor that calls a super constructor also
includes any non-static initializers of enclosing class. No
value is returned from a constructor execution join point, so its
return type is considered to be void.
</dd><dt><a name="d0e4567"></a><span class="term">Static initializer execution</span></dt><dd>
When the static initializer for a class executes. No value is
returned from a static initializer execution join point, so its
return type is considered to be void.
</dd><dt><a name="d0e4572"></a><span class="term">Object pre-initialization</span></dt><dd>
Before the object initialization code for a particular class runs.
This encompasses the time between the start of its first called
constructor and the start of its parent's constructor. Thus, the
execution of these join points encompass the join points of the
evaluation of the arguments of <tt>this()</tt> and
<tt>super()</tt> constructor calls. No value is
returned from an object pre-initialization join point, so its
return type is considered to be void.
</dd><dt><a name="d0e4583"></a><span class="term">Object initialization</span></dt><dd>
When the object initialization code for a particular class runs.
This encompasses the time between the return of its parent's
constructor and the return of its first called constructor. It
includes all the dynamic initializers and constructors used to
create the object. The object being constructed is the currently
executing object, and so may be accessed with the
<tt>this</tt> pointcut. No value is returned from a
constructor execution join point, so its return type is
considered to be void.
</dd><dt><a name="d0e4591"></a><span class="term">Field reference</span></dt><dd>
When a non-constant field is referenced. [Note that references
to constant fields (static final fields bound to a constant
string object or primitive value) are not join points, since Java
requires them to be inlined.]
</dd><dt><a name="d0e4596"></a><span class="term">Field set</span></dt><dd>
When a field is assigned to.
Field set join points are considered to have one argument,
the value the field is being set to.
No value is returned from a field set join point, so
its return type is considered to be void.
[Note that the initializations of constant fields (static
final fields where the initializer is a constant string object or
primitive value) are not join points, since Java requires their
references to be inlined.]
</dd><dt><a name="d0e4601"></a><span class="term">Handler execution</span></dt><dd>
When an exception handler executes.
Handler execution join points are considered to have one argument,
the exception being handled.
No value is returned from a field set join point, so
its return type is considered to be void.
</dd><dt><a name="d0e4606"></a><span class="term">Advice execution</span></dt><dd>
When the body of code for a piece of advice executes.
</dd></dl></div><p>
Each join point potentially has three pieces of state associated
with it: the currently executing object, the target object, and
an object array of arguments. These are exposed by the three
state-exposing pointcuts, <tt>this</tt>,
<tt>target</tt>, and <tt>args</tt>,
respectively.
</p><p>
Informally, the currently executing object is the object that a
<tt>this</tt> expression would pick out at the join
point. The target object is where control or attention is
transferred to by the join point. The arguments are those
values passed for that transfer of control or attention.
</p><div class="informaltable" id="d0e4627"><a name="d0e4627"></a><table border="1"><colgroup><col><col><col><col></colgroup><thead valign="top"><tr><th><span class="bold"><b>Join Point</b></span></th><th><span class="bold"><b>Current Object</b></span></th><th><span class="bold"><b>Target Object</b></span></th><th><span class="bold"><b>Arguments</b></span></th></tr></thead><tbody><tr><td>Method Call</td><td>executing object*</td><td>target object**</td><td>method arguments</td></tr><tr><td>Method Execution</td><td>executing object*</td><td>executing object*</td><td>method arguments</td></tr><tr><td>Constructor Call</td><td>executing object*</td><td>None</td><td>constructor arguments</td></tr><tr><td>Constructor Execution</td><td>executing object</td><td>executing object</td><td>constructor arguments</td></tr><tr><td>Static initializer execution</td><td>None</td><td>None</td><td>None</td></tr><tr><td>Object pre-initialization</td><td>None</td><td>None</td><td>constructor arguments</td></tr><tr><td>Object initialization</td><td>executing object</td><td>executing object</td><td>constructor arguments</td></tr><tr><td>Field reference</td><td>executing object*</td><td>target object**</td><td>None</td></tr><tr><td>Field assignment</td><td>executing object*</td><td>target object**</td><td>assigned value</td></tr><tr><td>Handler execution</td><td>executing object*</td><td>executing object*</td><td>caught exception</td></tr><tr><td>Advice execution</td><td>executing aspect</td><td>executing aspect</td><td>advice arguments</td></tr></tbody></table></div><p>* There is no executing object in static contexts such as
static method bodies or static initializers.
</p><p>** There is no target object for join points associated
with static methods or fields.
</p></div><div class="sect1"><a name="semantics-pointcuts"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="semantics-pointcuts"></a>Pointcuts</h2></div></div><p>
A pointcut is a program element that picks out join points and
exposes data from the execution context of those join points.
Pointcuts are used primarily by advice. They can be composed with
boolean operators to build up other pointcuts. The primitive
pointcuts and combinators provided by the language are:
</p><div class="variablelist"><dl><dt><a name="d0e4754"></a><span class="term"><tt>call(<i><tt>MethodPattern</tt></i>)</tt></span></dt><dd>
Picks out each method call join point whose signature matches
<i><tt>MethodPattern</tt></i>.
</dd><dt><a name="d0e4766"></a><span class="term"><tt>execution(<i><tt>MethodPattern</tt></i>)</tt></span></dt><dd>
Picks out each method execution join point whose signature matches
<i><tt>MethodPattern</tt></i>.
</dd><dt><a name="d0e4778"></a><span class="term"><tt>get(<i><tt>FieldPattern</tt></i>)</tt></span></dt><dd>
Picks out each field reference join point whose signature matches
<i><tt>FieldPattern</tt></i>.
[Note that references to constant fields (static final
fields bound to a constant string object or primitive value) are not
join points, since Java requires them to be inlined.]
</dd><dt><a name="d0e4790"></a><span class="term"><tt>set(<i><tt>FieldPattern</tt></i>)</tt></span></dt><dd>
Picks out each field set join point whose signature matches
<i><tt>FieldPattern</tt></i>.
[Note that the initializations of constant fields (static
final fields where the initializer is a constant string object or
primitive value) are not join points, since Java requires their
references to be inlined.]
</dd><dt><a name="d0e4802"></a><span class="term"><tt>call(<i><tt>ConstructorPattern</tt></i>)</tt></span></dt><dd>
Picks out each constructor call join point whose signature matches
<i><tt>ConstructorPattern</tt></i>.
</dd><dt><a name="d0e4814"></a><span class="term"><tt>execution(<i><tt>ConstructorPattern</tt></i>)</tt></span></dt><dd>
Picks out each constructor execution join point whose signature matches
<i><tt>ConstructorPattern</tt></i>.
</dd><dt><a name="d0e4826"></a><span class="term"><tt>initialization(<i><tt>ConstructorPattern</tt></i>)</tt></span></dt><dd>
Picks out each object initialization join point whose signature matches
<i><tt>ConstructorPattern</tt></i>.
</dd><dt><a name="d0e4838"></a><span class="term"><tt>preinitialization(<i><tt>ConstructorPattern</tt></i>)</tt></span></dt><dd>
Picks out each object pre-initialization join point whose signature matches
<i><tt>ConstructorPattern</tt></i>.
</dd><dt><a name="d0e4850"></a><span class="term"><tt>staticinitialization(<i><tt>TypePattern</tt></i>)</tt></span></dt><dd>
Picks out each static initializer execution join point whose signature matches
<i><tt>TypePattern</tt></i>.
</dd><dt><a name="d0e4862"></a><span class="term"><tt>handler(<i><tt>TypePattern</tt></i>)</tt></span></dt><dd>
Picks out each exception handler join point whose signature matches
<i><tt>TypePattern</tt></i>.
</dd><dt><a name="d0e4874"></a><span class="term"><tt>adviceexecution()</tt></span></dt><dd>
Picks out all advice execution join points.
</dd><dt><a name="d0e4880"></a><span class="term"><tt>within(<i><tt>TypePattern</tt></i>)</tt></span></dt><dd>
Picks out each join point where the executing code is defined
in a type matched by <i><tt>TypePattern</tt></i>.
</dd><dt><a name="d0e4892"></a><span class="term"><tt>withincode(<i><tt>MethodPattern</tt></i>)</tt></span></dt><dd>
Picks out each join point where the executing code is defined in
a method whose signature matches
<i><tt>MethodPattern</tt></i>.
</dd><dt><a name="d0e4904"></a><span class="term"><tt>withincode(<i><tt>ConstructorPattern</tt></i>)</tt></span></dt><dd>
Picks out each join point where the executing code is defined
in a constructor whose signature matches
<i><tt>ConstructorPattern</tt></i>.
</dd><dt><a name="d0e4916"></a><span class="term"><tt>cflow(<i><tt>Pointcut</tt></i>)</tt></span></dt><dd>
Picks out each join point in the control flow of any join point
<i><tt>P</tt></i> picked out by
<i><tt>Pointcut</tt></i>, including
<i><tt>P</tt></i> itself.
</dd><dt><a name="d0e4934"></a><span class="term"><tt>cflowbelow(<i><tt>Pointcut</tt></i>)</tt></span></dt><dd>
Picks out each join point in the control flow of any join point
<i><tt>P</tt></i> picked out by
<i><tt>Pointcut</tt></i>, but not
<i><tt>P</tt></i> itself.
</dd><dt><a name="d0e4952"></a><span class="term"><tt>this(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>)</tt></span></dt><dd>
Picks out each join point where the currently executing object
(the object bound to <tt>this</tt>) is an instance of
<i><tt>Type</tt></i>, or of the type of the
identifier <i><tt>Id</tt></i> (which must be bound in the enclosing
advice or pointcut definition).
Will not match any join points from static contexts.
</dd><dt><a name="d0e4973"></a><span class="term"><tt>target(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>)</tt></span></dt><dd>
Picks out each join point where the target object (the object
on which a call or field operation is applied to) is an instance of
<i><tt>Type</tt></i>, or of the type of the identifier
<i><tt>Id</tt></i> (which must be bound in the enclosing
advice or pointcut definition).
Will not match any calls, gets, or sets of static members.
</dd><dt><a name="d0e4991"></a><span class="term"><tt>args(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>, ...)</tt></span></dt><dd>
Picks out each join point where the arguments are instances of
the appropriate type (or type of the identifier if using that form). A
<tt>null</tt> argument is matched iff the static type of the
argument (declared parameter type or field type) is the same as, or a subtype of,
the specified args type.
</dd><dt><a name="d0e5006"></a><span class="term"><tt><i><tt>PointcutId</tt></i>(<i><tt>TypePattern</tt></i> or <i><tt>Id</tt></i>, ...)</tt></span></dt><dd>
Picks out each join point that is picked out by the
user-defined pointcut designator named by
<i><tt>PointcutId</tt></i>.
</dd><dt><a name="d0e5023"></a><span class="term"><tt>if(<i><tt>BooleanExpression</tt></i>)</tt></span></dt><dd>
Picks out each join point where the boolean expression
evaluates to <tt>true</tt>. The boolean expression used
can only access static members, parameters exposed by the enclosing
pointcut or advice, and <tt>thisJoinPoint</tt> forms. In
particular, it cannot call non-static methods on the aspect or
use return values or exceptions exposed by after advice.
</dd><dt><a name="d0e5038"></a><span class="term"><tt>! <i><tt>Pointcut</tt></i></tt></span></dt><dd>
Picks out each join point that is not picked out by
<i><tt>Pointcut</tt></i>.
</dd><dt><a name="d0e5049"></a><span class="term"><tt><i><tt>Pointcut0</tt></i> &amp;&amp; <i><tt>Pointcut1</tt></i></tt></span></dt><dd>
Picks out each join points that is picked out by both
<i><tt>Pointcut0</tt></i> and
<i><tt>Pointcut1</tt></i>.
</dd><dt><a name="d0e5065"></a><span class="term"><tt><i><tt>Pointcut0</tt></i> || <i><tt>Pointcut1</tt></i></tt></span></dt><dd>
Picks out each join point that is picked out by either
pointcuts. <i><tt>Pointcut0</tt></i> or
<i><tt>Pointcut1</tt></i>.
</dd><dt><a name="d0e5081"></a><span class="term"><tt>( <i><tt>Pointcut</tt></i> )</tt></span></dt><dd>
Picks out each join points picked out by
<i><tt>Pointcut</tt></i>.
</dd></dl></div><div class="sect2"><a name="pointcut-definition"></a><div class="titlepage"><div><h3 class="title"><a name="pointcut-definition"></a>Pointcut definition</h3></div></div><p>
Pointcuts are defined and named by the programmer with the
<tt>pointcut</tt> declaration.
</p><pre class="programlisting">
pointcut publicIntCall(int i):
call(public * *(int)) &amp;&amp; args(i);
</pre><p>
A named pointcut may be defined in either a class or aspect, and is
treated as a member of the class or aspect where it is found. As a
member, it may have an access modifier such as
<tt>public</tt> or <tt>private</tt>.
</p><pre class="programlisting">
class C {
pointcut publicCall(int i):
call(public * *(int)) &amp;&amp; args(i);
}
class D {
pointcut myPublicCall(int i):
C.publicCall(i) &amp;&amp; within(SomeType);
}
</pre><p>
Pointcuts that are not final may be declared abstract, and defined
without a body. Abstract pointcuts may only be declared within
abstract aspects.
</p><pre class="programlisting">
abstract aspect A {
abstract pointcut publicCall(int i);
}
</pre><p>
In such a case, an extending aspect may override the abstract
pointcut.
</p><pre class="programlisting">
aspect B extends A {
pointcut publicCall(int i): call(public Foo.m(int)) &amp;&amp; args(i);
}
</pre><p>
For completeness, a pointcut with a declaration may be declared
<tt>final</tt>.
</p><p>
Though named pointcut declarations appear somewhat like method
declarations, and can be overridden in subaspects, they cannot be
overloaded. It is an error for two pointcuts to be named with the
same name in the same class or aspect declaration.
</p><p>
The scope of a named pointcut is the enclosing class declaration.
This is different than the scope of other members; the scope of
other members is the enclosing class <span class="emphasis"><i>body</i></span>.
This means that the following code is legal:
</p><pre class="programlisting">
aspect B percflow(publicCall()) {
pointcut publicCall(): call(public Foo.m(int));
}
</pre></div><div class="sect2"><a name="context-exposure"></a><div class="titlepage"><div><h3 class="title"><a name="context-exposure"></a>Context exposure</h3></div></div><p>
Pointcuts have an interface; they expose some parts of the
execution context of the join points they pick out. For example,
the PublicIntCall above exposes the first argument from the
receptions of all public unary integer methods. This context is
exposed by providing typed formal parameters to named pointcuts and
advice, like the formal parameters of a Java method. These formal
parameters are bound by name matching.
</p><p>
On the right-hand side of advice or pointcut declarations, in
certain pointcut designators, a Java identifier is allowed in place
of a type or collection of types. The pointcut designators that
allow this are <tt>this</tt>, <tt>target</tt>,
and <tt>args</tt>. In all such cases, using an
identifier rather than a type does two things. First, it selects
join points as based on the type of the formal parameter. So the
pointcut
</p><pre class="programlisting">
pointcut intArg(int i): args(i);
</pre><p>
picks out join points where an <tt>int</tt> (or
a <tt>byte</tt>, <tt>short</tt>, or
<tt>char</tt>; anything assignable to an
<tt>int</tt>) is being passed as an argument.
Second, though, it makes the value of that argument
available to the enclosing advice or pointcut.
</p><p>
Values can be exposed from named pointcuts as well, so
</p><pre class="programlisting">
pointcut publicCall(int x): call(public *.*(int)) &amp;&amp; intArg(x);
pointcut intArg(int i): args(i);
</pre><p>
is a legal way to pick out all calls to public methods accepting an
int argument, and exposing that argument.
</p><p>
There is one special case for this kind of exposure. Exposing an
argument of type Object will also match primitive typed arguments,
and expose a "boxed" version of the primitive. So,
</p><pre class="programlisting">
pointcut publicCall(): call(public *.*(..)) &amp;&amp; args(Object);
</pre><p>
will pick out all unary methods that take, as their only argument,
subtypes of Object (i.e., not primitive types like
<tt>int</tt>), but
</p><pre class="programlisting">
pointcut publicCall(Object o): call(public *.*(..)) &amp;&amp; args(o);
</pre><p>
will pick out all unary methods that take any argument: And if the
argument was an <tt>int</tt>, then the value passed to
advice will be of type <tt>java.lang.Integer</tt>.
</p><p>
The "boxing" of the primitive value is based on the
<span class="emphasis"><i>original</i></span> primitive type. So in the
following program
</p><pre class="programlisting">
public class InstanceOf {
public static void main(String[] args) {
doInt(5);
}
static void doInt(int i) { }
}
aspect IntToLong {
pointcut el(long l) :
execution(* doInt(..)) &amp;&amp; args(l);
before(Object o) : el(o) {
System.out.println(o.getClass());
}
}
</pre><p>
The pointcut will match and expose the integer argument,
but it will expose it as an <tt>Integer</tt>,
not a <tt>Long</tt>.
</p></div><div class="sect2"><a name="primitive-pointcuts"></a><div class="titlepage"><div><h3 class="title"><a name="primitive-pointcuts"></a>Primitive pointcuts</h3></div></div><div class="sect3"><a name="d0e5213"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5213"></a>Method-related pointcuts</h4></div></div><p>AspectJ provides two primitive pointcut designators designed to
capture method call and execution join points. </p><div class="itemizedlist"><ul><li><a name="d0e5219"></a><tt>call(<i><tt>MethodPattern</tt></i>)</tt></li><li><a name="d0e5225"></a><tt>execution(<i><tt>MethodPattern</tt></i>)</tt></li></ul></div></div><div class="sect3"><a name="d0e5231"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5231"></a>Field-related pointcuts</h4></div></div><p>
AspectJ provides two primitive pointcut designators designed to
capture field reference and set join points:
</p><div class="itemizedlist"><ul><li><a name="d0e5237"></a><tt>get(<i><tt>FieldPattern</tt></i>)</tt></li><li><a name="d0e5243"></a><tt>set(<i><tt>FieldPattern</tt></i>)</tt></li></ul></div><p>
All set join points are treated as having one argument, the value the
field is being set to, so at a set join point, that value can be
accessed with an <tt>args</tt> pointcut. So an aspect
guarding a static integer variable x declared in type T might be written as
</p><pre class="programlisting">
aspect GuardedX {
static final int MAX_CHANGE = 100;
before(int newval): set(static int T.x) &amp;&amp; args(newval) {
if (Math.abs(newval - T.x) &gt; MAX_CHANGE)
throw new RuntimeException();
}
}
</pre></div><div class="sect3"><a name="d0e5256"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5256"></a>Object creation-related pointcuts</h4></div></div><p>
AspectJ provides primitive pointcut designators designed to
capture the initializer execution join points of objects.
</p><div class="itemizedlist"><ul><li><a name="d0e5262"></a><tt>call(<i><tt>ConstructorPattern</tt></i>)</tt></li><li><a name="d0e5268"></a><tt>execution(<i><tt>ConstructorPattern</tt></i>)</tt></li><li><a name="d0e5274"></a><tt>initialization(<i><tt>ConstructorPattern</tt></i>)</tt></li><li><a name="d0e5280"></a><tt>preinitialization(<i><tt>ConstructorPattern</tt></i>)</tt></li></ul></div></div><div class="sect3"><a name="d0e5286"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5286"></a>Class initialization-related pointcuts</h4></div></div><p>
AspectJ provides one primitive pointcut designator to pick out
static initializer execution join points.
</p><div class="itemizedlist"><ul><li><a name="d0e5292"></a><tt>staticinitialization(<i><tt>TypePattern</tt></i>)</tt></li></ul></div></div><div class="sect3"><a name="d0e5298"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5298"></a>Exception handler execution-related pointcuts</h4></div></div><p>
AspectJ provides one primitive pointcut designator to capture
execution of exception handlers:
</p><div class="itemizedlist"><ul><li><a name="d0e5304"></a><tt>handler(<i><tt>TypePattern</tt></i>)</tt></li></ul></div><p>
All handler join points are treated as having one argument, the value
of the exception being handled. That value can be accessed with an
<tt>args</tt> pointcut. So an aspect used to put
<tt>FooException</tt> objects into some normal form before
they are handled could be written as
</p><pre class="programlisting">
aspect NormalizeFooException {
before(FooException e): handler(FooException) &amp;&amp; args(e) {
e.normalize();
}
}
</pre></div><div class="sect3"><a name="d0e5320"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5320"></a>Advice execution-related pointcuts</h4></div></div><p>
AspectJ provides one primitive pointcut designator to capture
execution of advice
</p><div class="itemizedlist"><ul><li><a name="d0e5326"></a><tt>adviceexecution()</tt></li></ul></div><p>
This can be used, for example, to filter out any join point in the
control flow of advice from a particular aspect.
</p><pre class="programlisting">
aspect TraceStuff {
pointcut myAdvice(): adviceexecution() &amp;&amp; within(TraceStuff);
before(): call(* *(..)) &amp;&amp; !cflow(myAdvice) {
// do something
}
}
</pre></div><div class="sect3"><a name="d0e5333"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5333"></a>State-based pointcuts</h4></div></div><p>
Many concerns cut across the dynamic times when an object of a
particular type is executing, being operated on, or being passed
around. AspectJ provides primitive pointcuts that capture join
points at these times. These pointcuts use the dynamic types of
their objects to pick out join points. They may also be used to
expose the objects used for discrimination.
</p><div class="itemizedlist"><ul><li><a name="d0e5339"></a><tt>this(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>)</tt></li><li><a name="d0e5348"></a><tt>target(<i><tt>Type</tt></i> or <i><tt>Id</tt></i>)</tt></li></ul></div><p>
The <tt>this</tt> pointcut picks out each join point where
the currently executing object (the object bound to
<tt>this</tt>) is an instance of a particular type. The
<tt>target</tt> pointcut picks out each join point where
the target object (the object on which a method is called or a field
is accessed) is an instance of a particular type. Note that
<tt>target</tt> should be understood to be the object the
current join point is transfering control to. This means that the
target object is the same as the current object at a method execution
join point, for example, but may be different at a method call join
point.
</p><div class="itemizedlist"><ul><li><a name="d0e5372"></a><tt>args(<i><tt>Type</tt></i> or <i><tt>Id</tt></i> or "..", ...)</tt></li></ul></div><p>
The args pointcut picks out each join point where the arguments are
instances of some types. Each element in the comma-separated list is
one of four things. If it is a type name, then the argument in that
position must be an instance of that type. If it is an identifier,
then that identifier must be bound in the enclosing advice or
pointcut declaration, and so the argument in that position must be an
instance of the type of the identifier (or of any type if the
identifier is typed to Object). If it is the "*" wildcard, then any
argument will match, and if it is the special wildcard "..", then any
number of arguments will match, just like in signature patterns. So the
pointcut
</p><pre class="programlisting">
args(int, .., String)
</pre><p>
will pick out all join points where the first argument is an
<tt>int</tt> and the last is a <tt>String</tt>.
</p></div><div class="sect3"><a name="d0e5393"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5393"></a>Control flow-based pointcuts</h4></div></div><p>
Some concerns cut across the control flow of the program. The
<tt>cflow</tt> and <tt>cflowbelow</tt> primitive
pointcut designators capture join points based on control flow.
</p><div class="itemizedlist"><ul><li><a name="d0e5405"></a><tt>cflow(<i><tt>Pointcut</tt></i>)</tt></li><li><a name="d0e5411"></a><tt>cflowbelow(<i><tt>Pointcut</tt></i>)</tt></li></ul></div><p>
The <tt>cflow</tt> pointcut picks out all join points that
occur between entry and exit of each join point
<i><tt>P</tt></i> picked out by
<i><tt>Pointcut</tt></i>, including
<i><tt>P</tt></i> itself. Hence, it picks out the join
points <span class="emphasis"><i>in</i></span> the control flow of the join points
picked out by <i><tt>Pointcut</tt></i>.
</p><p>
The <tt>cflowbelow</tt> pointcut picks out all join points
that occur between entry and exit of each join point
<i><tt>P</tt></i> picked out by
<i><tt>Pointcut</tt></i>, but not including
<i><tt>P</tt></i> itself. Hence, it picks out the join
points <span class="emphasis"><i>below</i></span> the control flow of the join points
picked out by <i><tt>Pointcut</tt></i>.
</p><div class="sect4"><a name="d0e5457"></a><div class="titlepage"><div><h5 class="title"><a name="d0e5457"></a>Context exposure from control flows</h5></div></div><p>
The <tt>cflow</tt> and
<tt>cflowbelow</tt> pointcuts may expose context
state through enclosed <tt>this</tt>,
<tt>target</tt>, and <tt>args</tt>
pointcuts.
</p><p>
Anytime such state is accessed, it is accessed through the
<span class="emphasis"><i>most recent</i></span> control flow that
matched. So the "current arg" that would be printed by
the following program is zero, even though it is in many
control flows.
</p><pre class="programlisting">
class Test {
public static void main(String[] args) {
fact(5);
}
static int fact(int x) {
if (x == 0) {
System.err.println("bottoming out");
return 1;
}
else return x * fact(x - 1);
}
}
aspect A {
pointcut entry(int i): call(int fact(int)) &amp;&amp; args(i);
pointcut writing(): call(void println(String)) &amp;&amp; ! within(A);
before(int i): writing() &amp;&amp; cflow(entry(i)) {
System.err.println("Current arg is " + i);
}
}
</pre><p>
It is an error to expose such state through
<span class="emphasis"><i>negated</i></span> control flow pointcuts, such
as within <tt>!
cflowbelow(<i><tt>P</tt></i>)</tt>.
</p></div></div><div class="sect3"><a name="d0e5495"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5495"></a>Program text-based pointcuts</h4></div></div><p>
While many concerns cut across the runtime structure of the program,
some must deal with the lexical structure. AspectJ allows aspects to
pick out join points based on where their associated code is defined.
</p><div class="itemizedlist"><ul><li><a name="d0e5501"></a><tt>within(<i><tt>TypePattern</tt></i>)</tt></li><li><a name="d0e5507"></a><tt>withincode(<i><tt>MethodPattern</tt></i>)</tt></li><li><a name="d0e5513"></a><tt>withincode(<i><tt>ConstructorPattern</tt></i>)</tt></li></ul></div><p>
The <tt>within</tt> pointcut picks out each join point
where the code executing is defined in the declaration of one of the
types in <i><tt>TypePattern</tt></i>. This includes the
class initialization, object initialization, and method and
constructor execution join points for the type, as well as any join
points associated with the statements and expressions of the type.
It also includes any join points that are associated with code in a
type's nested types, and that type's default constructor, if there is
one.
</p><p>
The <tt>withincode</tt> pointcuts picks out each join point
where the code executing is defined in the declaration of a
particular method or constructor. This includes the method or
constructor execution join point as well as any join points
associated with the statements and expressions of the method or
constructor. It also includes any join points that are associated
with code in a method or constructor's local or anonymous types.
</p></div><div class="sect3"><a name="d0e5532"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5532"></a>Expression-based pointcuts</h4></div></div><div class="itemizedlist"><ul><li><a name="d0e5536"></a><tt>if(<i><tt>BooleanExpression</tt></i>)</tt></li></ul></div><p>
The if pointcut picks out join points based on a dynamic property.
It's syntax takes an expression, which must evaluate to a boolean
true or false. Within this expression, the
<tt>thisJoinPoint</tt> object is available. So one
(extremely inefficient) way of picking out all call join points would
be to use the pointcut
</p><pre class="programlisting">
if(thisJoinPoint.getKind().equals("call"))
</pre><p>
Note that the order of evaluation for pointcut expression
components at a join point is undefined. Writing <tt>if</tt>
pointcuts that have side-effects is considered bad style and may also
lead to potentially confusing or even changing behavior with regard
to when or if the test code will run.
</p></div></div><div class="sect2"><a name="signatures"></a><div class="titlepage"><div><h3 class="title"><a name="signatures"></a>Signatures</h3></div></div><p>
One very important property of a join point is its signature, which is
used by many of AspectJ's pointcut designators to select particular
join points.
</p><div class="sect3"><a name="d0e5559"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5559"></a>Methods</h4></div></div><p>
Join points associated with methods typically have method signatures,
consisting of a method name, parameter types, return type, the types of
the declared (checked) exceptions, and some type that the method could
be called on (below called the "qualifying type").
</p><p>
At a method call join point, the signature is a method signature whose
qualifying type is the static type used to <span class="emphasis"><i>access</i></span>
the method. This means that the signature for the join point created
from the call <tt>((Integer)i).toString()</tt> is different
than that for the call <tt>((Object)i).toString()</tt>, even
if <tt>i</tt> is the same variable.
</p><p>
At a method execution join point, the signature is a method signature
whose qualifying type is the declaring type of the method.
</p></div><div class="sect3"><a name="d0e5580"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5580"></a>Fields</h4></div></div><p>
Join points associated with fields typically have field signatures,
consisting of a field name and a field type. A field reference join
point has such a signature, and no parameters. A field set join point
has such a signature, but has a has a single parameter whose type is
the same as the field type.
</p></div><div class="sect3"><a name="d0e5585"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5585"></a>Constructors</h4></div></div><p>
Join points associated with constructors typically have constructor
signatures, consisting of a parameter types, the types of the declared
(checked) exceptions, and the declaring type.
</p><p>
At a constructor call join point, the signature is the constructor
signature of the called constructor. At a constructor execution join
point, the signature is the constructor signature of the currently
executing constructor.
</p><p>
At object initialization and pre-initialization join points, the
signature is the constructor signature for the constructor that started
this initialization: the first constructor entered during this type's
initialization of this object.
</p></div><div class="sect3"><a name="d0e5594"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5594"></a>Others</h4></div></div><p>
At a handler execution join point, the signature is composed of the
exception type that the handler handles.
</p><p>
At an advice execution join point, the signature is composed of the
aspect type, the parameter types of the advice, the return type (void
for all but around advice) and the types of the declared (checked)
exceptions.
</p></div></div><div class="sect2"><a name="matching"></a><div class="titlepage"><div><h3 class="title"><a name="matching"></a>Matching</h3></div></div><p>
The <tt>withincode</tt>, <tt>call</tt>,
<tt>execution</tt>, <tt>get</tt>, and
<tt>set</tt> primitive pointcut designators all use signature
patterns to determine the join points they describe. A signature
pattern is an abstract description of one or more join-point
signatures. Signature patterns are intended to match very closely the
same kind of things one would write when declaring individual members
and constructors.
</p><p>
Method declarations in Java include method names, method parameters,
return types, modifiers like static or private, and throws clauses,
while constructor declarations omit the return type and replace the
method name with the class name. The start of a particular method
declaration, in class <tt>Test</tt>, for example, might be
</p><pre class="programlisting">
class C {
public final void foo() throws ArrayOutOfBoundsException { ... }
}
</pre><p>
In AspectJ, method signature patterns have all these, but most elements
can be replaced by wildcards. So
</p><pre class="programlisting">
call(public final void C.foo() throws ArrayOutOfBoundsException)
</pre><p>
picks out call join points to that method, and the pointcut
</p><pre class="programlisting">
call(public final void *.*() throws ArrayOutOfBoundsException)
</pre><p>
picks out all call join points to methods, regardless of their name
name or which class they are defined on, so long as they take no
arguments, return no value, are both <tt>public</tt> and
<tt>final</tt>, and are declared to throw
<tt>ArrayOutOfBounds</tt> exceptions.
</p><p>
The defining type name, if not present, defaults to *, so another way
of writing that pointcut would be
</p><pre class="programlisting">
call(public final void *() throws ArrayOutOfBoundsException)
</pre><p>
The wildcard <tt>..</tt> indicates zero or more
parameters, so
</p><pre class="programlisting">
execution(void m(..))
</pre><p>
picks out execution join points for void methods named
<tt>m</tt>, of any number of arguments, while
</p><pre class="programlisting">
execution(void m(.., int))
</pre><p>
picks out execution join points for void methods named
<tt>m</tt> whose last parameter is of type
<tt>int</tt>.
</p><p>
The modifiers also form part of the signature pattern. If an AspectJ
signature pattern should match methods without a particular modifier,
such as all non-public methods, the appropriate modifier should be
negated with the <tt>!</tt> operator. So,
</p><pre class="programlisting">
withincode(!public void foo())
</pre><p>
picks out all join points associated with code in null non-public
void methods named <tt>foo</tt>, while
</p><pre class="programlisting">
withincode(void foo())
</pre><p>
picks out all join points associated with code in null void methods
named <tt>foo</tt>, regardless of access modifier.
</p><p>
Method names may contain the * wildcard, indicating any number of
characters in the method name. So
</p><pre class="programlisting">
call(int *())
</pre><p>
picks out all call join points to <tt>int</tt> methods
regardless of name, but
</p><pre class="programlisting">
call(int get*())
</pre><p>
picks out all call join points to <tt>int</tt> methods
where the method name starts with the characters "get".
</p><p>
AspectJ uses the <tt>new</tt> keyword for constructor
signature patterns rather than using a particular class name. So the
execution join points of private null constructor of a class C
defined to throw an ArithmeticException can be picked out with
</p><pre class="programlisting">
execution(private C.new() throws ArithmeticException)
</pre><div class="sect3"><a name="d0e5716"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5716"></a>Matching based on the declaring type</h4></div></div><p>
The signature-matching pointcuts all specify a declaring type,
but the meaning varies slightly for each join point signature,
in line with Java semantics.
</p><p>
When matching for pointcuts <tt>withincode</tt>,
<tt>get</tt>, and <tt>set</tt>, the declaring
type is the class that contains the declaration.
</p><p>
When matching method-call join points, the
declaring type is the static type used to access the method.
A common mistake is to specify a declaring type for the
<tt>call</tt> pointcut that is a subtype of the
originally-declaring type. For example, given the class
</p><pre class="programlisting">
class Service implements Runnable {
public void run() { ... }
}
</pre><p>
the following pointcut
</p><pre class="programlisting">
call(void Service.run())
</pre><p>
would fail to pick out the join point for the code
</p><pre class="programlisting">
((Runnable) new Service()).run();
</pre><p>
Specifying the originally-declaring type is correct, but would
pick out any such call (here, calls to the <tt>run()</tt>
method of any Runnable).
In this situation, consider instead picking out the target type:
</p><pre class="programlisting">
call(void run()) &amp;&amp; target(Service)
</pre><p>
When matching method-execution join points,
if the execution pointcut method signature specifies a declaring type,
the pointcut will only match methods declared in that type, or methods
that override methods declared in or inherited by that type.
So the pointcut
</p><pre class="programlisting">
execution(public void Middle.*())
</pre><p>
picks out all method executions for public methods returning void
and having no arguments that are either declared in, or inherited by,
Middle, even if those methods are overridden in a subclass of Middle.
So the pointcut would pick out the method-execution join point
for Sub.m() in this code:
</p><pre class="programlisting">
class Super {
protected void m() { ... }
}
class Middle extends Super {
}
class Sub extends Middle {
public void m() { ... }
}
</pre></div><div class="sect3"><a name="d0e5762"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5762"></a>Matching based on the throws clause</h4></div></div><p>
Type patterns may be used to pick out methods and constructors
based on their throws clauses. This allows the following two
kinds of extremely wildcarded pointcuts:
</p><pre class="programlisting">
pointcut throwsMathlike():
// each call to a method with a throws clause containing at least
// one exception exception with "Math" in its name.
call(* *(..) throws *..*Math*);
pointcut doesNotThrowMathlike():
// each call to a method with a throws clause containing no
// exceptions with "Math" in its name.
call(* *(..) throws !*..*Math*);
</pre><p>
A <i><tt>ThrowsClausePattern</tt></i> is a comma-separated list of
<i><tt>ThrowsClausePatternItem</tt></i>s, where
<div class="variablelist"><dl><dt><a name="d0e5778"></a><span class="term"><i><tt>ThrowsClausePatternItem</tt></i> :</span></dt><dd><tt>[ ! ]
<i><tt>TypeNamePattern</tt></i></tt></dd></dl></div>
</p><p>
A <i><tt>ThrowsClausePattern</tt></i> matches the
throws clause of any code member signature. To match, each
<tt>ThrowsClausePatternItem</tt> must
match the throws clause of the member in question. If any item
doesn't match, then the whole pattern doesn't match.
</p><p>
If a ThrowsClausePatternItem begins with "!", then it matches a
particular throws clause if and only if <span class="emphasis"><i>none</i></span>
of the types named in the throws clause is matched by the
<tt>TypeNamePattern</tt>.
</p><p>
If a <i><tt>ThrowsClausePatternItem</tt></i> does not
begin with "!", then it matches a throws clause if and only if
<span class="emphasis"><i>any</i></span> of the types named in the throws clause
is matched by the <span class="emphasis"><i>TypeNamePattern</i></span>.
</p><p>
The rule for "!" matching has one potentially surprising
property, in that these two pointcuts
<div class="itemizedlist"><ul><li><a name="d0e5819"></a> call(* *(..) throws !IOException) </li><li><a name="d0e5821"></a> call(* *(..) throws (!IOException)) </li></ul></div>
will match differently on calls to
<blockquote class="blockquote"><tt>
void m() throws RuntimeException, IOException {}
</tt></blockquote>
</p><p>
[1] will NOT match the method m(), because method m's throws
clause declares that it throws IOException. [2] WILL match the
method m(), because method m's throws clause declares the it
throws some exception which does not match IOException,
i.e. RuntimeException.
</p></div></div><div class="sect2"><a name="type-patterns"></a><div class="titlepage"><div><h3 class="title"><a name="type-patterns"></a>Type patterns</h3></div></div><p>
Type patterns are a way to pick out collections of types and use them
in places where you would otherwise use only one type. The rules for
using type patterns are simple.
</p><div class="sect3"><a name="d0e5835"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5835"></a>Exact type pattern</h4></div></div><p>
First, all type names are also type patterns. So
<tt>Object</tt>, <tt>java.util.HashMap</tt>,
<tt>Map.Entry</tt>, <tt>int</tt> are all type
patterns.
</p><p>
If a type pattern is an exact type - if it doesn't
include a wildcard - then the matching works just
like normal type lookup in Java: </p><div class="itemizedlist"><ul><li><a name="d0e5855"></a>Patterns that have the same names as
primitive types (like <tt>int</tt>) match
those primitive types.</li><li><a name="d0e5860"></a>Patterns that are qualified by package names
(like <tt>java.util.HashMap</tt>) match types
in other packages.
</li><li><a name="d0e5865"></a>Patterns that are not qualified (like
<tt>HashMap</tt>) match types that are
resolved by Java's normal scope rules. So, for
example, <tt>HashMap</tt> might match a
package-level type in the same package or a type that
have been imported with java's
<tt>import</tt> form. But it would not match
<tt>java.util.HashMap</tt> unless the aspect
were in <tt>java.util</tt> or the type had
been imported.
</li></ul></div><p>
So exact type patterns match based on usual Java scope
rules.
</p></div><div class="sect3"><a name="d0e5884"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5884"></a>Type name patterns</h4></div></div><p>
There is a special type name, *, which is also a type pattern. * picks out all
types, including primitive types. So
</p><pre class="programlisting">
call(void foo(*))
</pre><p>
picks out all call join points to void methods named foo, taking one
argument of any type.
</p><p>
Type names that contain the two wildcards "*" and
"<tt>..</tt>" are also type patterns. The * wildcard matches
zero or more characters characters except for ".", so it can be used
when types have a certain naming convention. So
</p><pre class="programlisting">
handler(java.util.*Map)
</pre><p>
picks out the types java.util.Map and java.util.java.util.HashMap,
among others, and
</p><pre class="programlisting">
handler(java.util.*)
</pre><p>
picks out all types that start with "<tt>java.util.</tt>" and
don't have any more "."s, that is, the types in the
<tt>java.util</tt> package, but not inner types
(such as java.util.Map.Entry).
</p><p>
The "<tt>..</tt>" wildcard matches any sequence of
characters that start and end with a ".", so it can be used
to pick out all types in any subpackage, or all inner types. So
</p><pre class="programlisting">
within(com.xerox..*)
</pre><p>
picks out all join points where the code is in any
declaration of a type whose name begins with "<tt>com.xerox.</tt>".
</p><p>
Type patterns with wildcards do not depend on Java's
usual scope rules - they match against all types
available to the weaver, not just those that are
imported into an Aspect's declaring file.
</p></div><div class="sect3"><a name="d0e5926"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5926"></a>Subtype patterns</h4></div></div><p>
It is possible to pick out all subtypes of a type (or a collection of
types) with the "+" wildcard. The "+" wildcard follows immediately a
type name pattern. So, while
</p><pre class="programlisting">
call(Foo.new())
</pre><p>
picks out all constructor call join points where an instance of exactly
type Foo is constructed,
</p><pre class="programlisting">
call(Foo+.new())
</pre><p>
picks out all constructor call join points where an instance of any
subtype of Foo (including Foo itself) is constructed, and the unlikely
</p><pre class="programlisting">
call(*Handler+.new())
</pre><p>
picks out all constructor call join points where an instance of any
subtype of any type whose name ends in "Handler" is constructed.
</p></div><div class="sect3"><a name="d0e5943"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5943"></a>Array type patterns</h4></div></div><p>
A type name pattern or subtype pattern can be followed by one or more
sets of square brackets to make array type patterns. So
<tt>Object[]</tt> is an array type pattern, and so is
<tt>com.xerox..*[][]</tt>, and so is
<tt>Object+[]</tt>.
</p></div><div class="sect3"><a name="d0e5957"></a><div class="titlepage"><div><h4 class="title"><a name="d0e5957"></a>Type patterns</h4></div></div><p>
Type patterns are built up out of type name patterns, subtype patterns,
and array type patterns, and constructed with boolean operators
<tt>&amp;&amp;</tt>, <tt>||</tt>, and
<tt>!</tt>. So
</p><pre class="programlisting">
staticinitialization(Foo || Bar)
</pre><p>
picks out the static initializer execution join points of either Foo or Bar,
and
</p><pre class="programlisting">
call((Foo+ &amp;&amp; ! Foo).new(..))
</pre><p>
picks out the constructor call join points when a subtype of Foo, but
not Foo itself, is constructed.
</p></div></div><div class="sect2"><a name="pattern-summary"></a><div class="titlepage"><div><h3 class="title"><a name="pattern-summary"></a>Pattern Summary</h3></div></div><p>
Here is a summary of the pattern syntax used in AspectJ:
</p><pre class="programlisting">
MethodPattern =
[ModifiersPattern] TypePattern
[TypePattern . ] IdPattern (TypePattern | ".." , ... )
[ throws ThrowsPattern ]
ConstructorPattern =
[ModifiersPattern ]
[TypePattern . ] new (TypePattern | ".." , ...)
[ throws ThrowsPattern ]
FieldPattern =
[ModifiersPattern] TypePattern [TypePattern . ] IdPattern
ThrowsPattern =
[ ! ] TypePattern , ...
TypePattern =
IdPattern [ + ] [ [] ... ]
| ! TypePattern
| TypePattern &amp;&amp; TypePattern
| TypePattern || TypePattern
| ( TypePattern )
IdPattern =
Sequence of characters, possibly with special * and .. wildcards
ModifiersPattern =
[ ! ] JavaModifier ...
</pre></div></div><div class="sect1"><a name="semantics-advice"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="semantics-advice"></a>Advice</h2></div></div><p>
Each piece of advice is of the form
<blockquote class="blockquote"><tt>[ strictfp ] <i><tt>AdviceSpec</tt></i> [
throws <i><tt>TypeList</tt></i> ] :
<i><tt>Pointcut</tt></i> {
<i><tt>Body</tt></i> } </tt></blockquote>
where <i><tt>AdviceSpec</tt></i> is one of
</p><div class="itemizedlist"><ul><li><a name="d0e6012"></a><tt>before( <i><tt>Formals</tt></i> ) </tt></li><li><a name="d0e6018"></a><tt>after( <i><tt>Formals</tt></i> ) returning
[ ( <i><tt>Formal</tt></i> ) ] </tt></li><li><a name="d0e6027"></a><tt>after( <i><tt>Formals</tt></i> ) throwing [
( <i><tt>Formal</tt></i> ) ] </tt></li><li><a name="d0e6036"></a><tt>after( <i><tt>Formals</tt></i> ) </tt></li><li><a name="d0e6042"></a><tt><i><tt>Type</tt></i>
around( <i><tt>Formals</tt></i> )</tt></li></ul></div><p>
and where <i><tt>Formal</tt></i> refers to a
variable binding like those used for method parameters,
of the form
<tt><i><tt>Type</tt></i></tt>
<tt><i><tt>Variable-Name</tt></i></tt>,
and <i><tt>Formals</tt></i> refers to a comma-delimited
list of <i><tt>Formal</tt></i>.
</p><p>
Advice defines crosscutting behavior. It is defined in terms of
pointcuts. The code of a piece of advice runs at every join point
picked out by its pointcut. Exactly how the code runs depends on the
kind of advice.
</p><p>
AspectJ supports three kinds of advice. The kind of advice determines how
it interacts with the join points it is defined over. Thus AspectJ
divides advice into that which runs before its join points, that which
runs after its join points, and that which runs in place of (or "around")
its join points.
</p><p>
While before advice is relatively unproblematic, there can be three
interpretations of after advice: After the execution of a join point
completes normally, after it throws an exception, or after it does either
one. AspectJ allows after advice for any of these situations.
</p><pre class="programlisting">
aspect A {
pointcut publicCall(): call(public Object *(..));
after() returning (Object o): publicCall() {
System.out.println("Returned normally with " + o);
}
after() throwing (Exception e): publicCall() {
System.out.println("Threw an exception: " + e);
}
after(): publicCall(){
System.out.println("Returned or threw an Exception");
}
}
</pre><p>
After returning advice may not care about its returned object, in which
case it may be written
</p><pre class="programlisting">
after() returning: call(public Object *(..)) {
System.out.println("Returned normally");
}
</pre><p>
If after returning does expose its returned object, then the
type of the parameter is considered to be an
<tt>instanceof</tt>-like constraint on the advice: it
will run only when the return value is of the appropriate type.
</p><p>
A value is of the appropriate type if it would be assignable to
a variable of that type, in the Java sense. That is, a
<tt>byte</tt> value is assignable to a
<tt>short</tt> parameter but not vice-versa, an
<tt>int</tt> is assignable to a
<tt>float</tt> parameter, <tt>boolean</tt>
values are only assignable to <tt>boolean</tt>
parameters, and reference types work by instanceof.
</p><p>
There are two special cases: If the exposed value is typed to
<tt>Object</tt>, then the advice is not constrained by
that type: the actual return value is converted to an object
type for the body of the advice: <tt>int</tt> values
are represented as <tt>java.lang.Integer</tt> objects,
etc, and no value (from void methods, for example) is
represented as <tt>null</tt>.
</p><p>
Secondly, the <tt>null</tt> value is assignable to a
parameter <tt>T</tt> if the join point
<span class="emphasis"><i>could</i></span> return something of type
<tt>T</tt>.
</p><p>
Around advice runs in place of the join point it operates over, rather
than before or after it. Because around is allowed to return a value, it
must be declared with a return type, like a method.
</p><p>
Thus, a simple use of around advice is to make a particular method
constant:
</p><pre class="programlisting">
aspect A {
int around(): call(int C.foo()) {
return 3;
}
}
</pre><p>
Within the body of around advice, though, the computation of the original
join point can be executed with the special syntax
</p><pre class="programlisting">
proceed( ... )
</pre><p>
The proceed form takes as arguments the context exposed by the around's
pointcut, and returns whatever the around is declared to return. So the
following around advice will double the second argument to
<tt>foo</tt> whenever it is called, and then halve its result:
</p><pre class="programlisting">
aspect A {
int around(int i): call(int C.foo(Object, int)) &amp;&amp; args(i) {
int newi = proceed(i*2)
return newi/2;
}
}
</pre><p>
If the return value of around advice is typed to
<tt>Object</tt>, then the result of proceed is converted to an
object representation, even if it is originally a primitive value. And
when the advice returns an Object value, that value is converted back to
whatever representation it was originally. So another way to write the
doubling and halving advice is:
</p><pre class="programlisting">
aspect A {
Object around(int i): call(int C.foo(Object, int)) &amp;&amp; args(i) {
Integer newi = (Integer) proceed(i*2)
return new Integer(newi.intValue() / 2);
}
}
</pre><p>
Any occurence of <tt>proceed(..)</tt> within the body of
around advice is treated as the special proceed form (even if the
aspect defines a method named <tt>proceed</tt>) unless a
target other than the aspect instance is specified as the recipient of
the call.
For example, in the following program the first
call to proceed will be treated as a method call to
the <tt>ICanProceed</tt> instance, whereas the second call to
proceed is treated as the special proceed form.
</p><pre class="programlisting">
aspect A {
Object around(ICanProceed canProceed) : execution(* *(..)) &amp;&amp; this(canProceed) {
canProceed.proceed(); // a method call
return proceed(canProceed); // the special proceed form
}
private Object proceed(ICanProceed canProceed) {
// this method cannot be called from inside the body of around advice in
// the aspect
}
}
</pre><p>
In all kinds of advice, the parameters of the advice behave exactly like
method parameters. In particular, assigning to any parameter affects
only the value of the parameter, not the value that it came from. This
means that
</p><pre class="programlisting">
aspect A {
after() returning (int i): call(int C.foo()) {
i = i * 2;
}
}
</pre><p>
will <span class="emphasis"><i>not</i></span> double the returned value of the advice.
Rather, it will double the local parameter. Changing the values of
parameters or return values of join points can be done by using around
advice.
</p><div class="sect2"><a name="advice-modifiers"></a><div class="titlepage"><div><h3 class="title"><a name="advice-modifiers"></a>Advice modifiers</h3></div></div><p>
The <tt>strictfp</tt> modifier is the only modifier allowed
on advice, and it has the effect of making all floating-point
expressions within the advice be FP-strict.
</p></div><div class="sect2"><a name="advice-and-checked-exceptions"></a><div class="titlepage"><div><h3 class="title"><a name="advice-and-checked-exceptions"></a>Advice and checked exceptions</h3></div></div><p>
An advice declaration must include a <tt>throws</tt> clause
listing the checked exceptions the body may throw. This list of
checked exceptions must be compatible with each target join point
of the advice, or an error is signalled by the compiler.
</p><p>
For example, in the following declarations:
</p><pre class="programlisting">
import java.io.FileNotFoundException;
class C {
int i;
int getI() { return i; }
}
aspect A {
before(): get(int C.i) {
throw new FileNotFoundException();
}
before() throws FileNotFoundException: get(int C.i) {
throw new FileNotFoundException();
}
}
</pre><p>
both pieces of advice are illegal. The first because the body throws
an undeclared checked exception, and the second because field get join
points cannot throw <tt>FileNotFoundException</tt>s.
</p><p> The exceptions that each kind of join point in AspectJ may throw are:
</p><div class="variablelist"><dl><dt><a name="d0e6208"></a><span class="term">method call and execution</span></dt><dd>
the checked exceptions declared by the target method's
<tt>throws</tt> clause.
</dd><dt><a name="d0e6216"></a><span class="term">constructor call and execution</span></dt><dd>
the checked exceptions declared by the target constructor's
<tt>throws</tt> clause.
</dd><dt><a name="d0e6224"></a><span class="term">field get and set</span></dt><dd>
no checked exceptions can be thrown from these join points.
</dd><dt><a name="d0e6229"></a><span class="term">exception handler execution</span></dt><dd>
the exceptions that can be thrown by the target exception handler.
</dd><dt><a name="d0e6234"></a><span class="term">static initializer execution</span></dt><dd>
no checked exceptions can be thrown from these join points.
</dd><dt><a name="d0e6239"></a><span class="term">pre-initialization and initialization</span></dt><dd>
any exception that is in the throws clause of
<span class="emphasis"><i>all</i></span> constructors of the initialized class.
</dd><dt><a name="d0e6247"></a><span class="term">advice execution</span></dt><dd>
any exception that is in the throws clause of the advice.
</dd></dl></div></div><div class="sect2"><a name="advice-precedence"></a><div class="titlepage"><div><h3 class="title"><a name="advice-precedence"></a>Advice precedence</h3></div></div><p>
Multiple pieces of advice may apply to the same join point. In such
cases, the resolution order of the advice is based on advice
precedence.
</p><div class="sect3"><a name="d0e6257"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6257"></a>Determining precedence</h4></div></div><p>There are a number of rules that determine whether a particular
piece of advice has precedence over another when they advise the same
join point. </p><p>If the two pieces of advice are defined in different aspects,
then there are three cases: </p><div class="itemizedlist"><ul><li><a name="d0e6265"></a>If aspect A is matched earlier than aspect B in some
<tt>declare precedence</tt> form, then all advice in
concrete aspect A has precedence over all advice in concrete aspect B
when they are on the same join point. </li><li><a name="d0e6270"></a>
Otherwise, if aspect A is a subaspect of aspect B, then all advice
defined in A has precedence over all advice defined in
B. So, unless otherwise specified with
<tt>declare precedence</tt>, advice in a subaspect
has precedence over advice in a superaspect.
</li><li><a name="d0e6275"></a>
Otherwise, if two pieces of advice are defined in two different
aspects, it is undefined which one has precedence.
</li></ul></div><p>If the two pieces of advice are defined in the same aspect, then
there are two cases: </p><div class="itemizedlist"><ul><li><a name="d0e6280"></a>If either are <tt>after</tt> advice, then the one that
appears later in the aspect has precedence over the one that appears
earlier. </li><li><a name="d0e6285"></a>Otherwise, then the one that appears earlier in the aspect
has precedence over the one that appears later.
</li></ul></div><p>These rules can lead to circularity, such as</p><pre class="programlisting">
aspect A {
before(): execution(void main(String[] args)) {}
after(): execution(void main(String[] args)) {}
before(): execution(void main(String[] args)) {}
}
</pre><p>such circularities will result in errors signalled by the compiler. </p></div><div class="sect3"><a name="d0e6293"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6293"></a>Effects of precedence</h4></div></div><p>At a particular join point, advice is ordered by precedence.</p><p>A piece of <tt>around</tt> advice controls whether
advice of lower precedence will run by calling
<tt>proceed</tt>. The call to <tt>proceed</tt>
will run the advice with next precedence, or the computation under the
join point if there is no further advice. </p><p>A piece of <tt>before</tt> advice can prevent advice of
lower precedence from running by throwing an exception. If it returns
normally, however, then the advice of the next precedence, or the
computation under the join pint if there is no further advice, will run.
</p><p>Running <tt>after returning</tt> advice will run the
advice of next precedence, or the computation under the join point if
there is no further advice. Then, if that computation returned
normally, the body of the advice will run. </p><p>Running <tt>after throwing</tt> advice will run the
advice of next precedence, or the computation under the join
point if there is no further advice. Then, if that computation threw
an exception of an appropriate type, the body of the advice will
run. </p><p>Running <tt>after</tt> advice will run the advice of
next precedence, or the computation under the join point if
there is no further advice. Then the body of the advice will
run. </p></div></div><div class="sect2"><a name="reflective-access-to-the-join-point"></a><div class="titlepage"><div><h3 class="title"><a name="reflective-access-to-the-join-point"></a>Reflective access to the join point</h3></div></div><p>
Three special variables are visible within bodies of advice
and within <tt>if()</tt> pointcut expressions:
<tt>thisJoinPoint</tt>,
<tt>thisJoinPointStaticPart</tt>, and
<tt>thisEnclosingJoinPointStaticPart</tt>. Each is bound to
an object that encapsulates some of the context of the advice's current
or enclosing join point. These variables exist because some pointcuts
may pick out very large collections of join points. For example, the
pointcut
</p><pre class="programlisting">
pointcut publicCall(): call(public * *(..));
</pre><p>
picks out calls to many methods. Yet the body of advice over this
pointcut may wish to have access to the method name or parameters of a
particular join point.
</p><p>
<tt>thisJoinPoint</tt> is bound to a complete join point
object.
</p><p>
<tt>thisJoinPointStaticPart</tt> is bound to a part of the
join point object that includes less information, but for which no
memory allocation is required on each execution of the advice. It is
equivalent to <tt>thisJoinPoint.getStaticPart()</tt>.
</p><p>
<tt>thisEnclosingJoinPointStaticPart</tt> is bound to the
static part of the join point enclosing the current join point. Only
the static part of this enclosing join point is available through this
mechanism.
</p><p>
Standard Java reflection uses objects from the
<tt>java.lang.reflect</tt> hierarchy to build up its
reflective objects. Similarly, AspectJ join point objects have types
in a type hierarchy. The type of objects bound to
<tt>thisJoinPoint</tt> is
<tt>org.aspectj.lang.JoinPoint</tt>, while
<tt>thisStaticJoinPoint</tt> is bound to objects of interface
type <tt>org.aspectj.lang.JoinPoint.StaticPart</tt>.
</p></div></div><div class="sect1"><a name="semantics-declare"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="semantics-declare"></a>Static crosscutting</h2></div></div><p>
Advice declarations change the behavior of classes they crosscut, but do
not change their static type structure. For crosscutting concerns that do
operate over the static structure of type hierarchies, AspectJ provides
inter-type member declarations and other <tt>declare</tt> forms.
</p><div class="sect2"><a name="inter-type-member-declarations"></a><div class="titlepage"><div><h3 class="title"><a name="inter-type-member-declarations"></a>Inter-type member declarations</h3></div></div><p>
AspectJ allows the declaration of members by aspects that are
associated with other types.
</p><p>
An inter-type method declaration looks like
</p><div class="itemizedlist"><ul><li><a name="d0e6401"></a><tt>
[ <i><tt>Modifiers</tt></i> ]
<i><tt>Type</tt></i> <i><tt>OnType</tt></i>
.
<i><tt>Id</tt></i>(<i><tt>Formals</tt></i>)
[ <i><tt>ThrowsClause</tt></i> ]
{ <i><tt>Body</tt></i> }</tt></li><li><a name="d0e6425"></a><tt>abstract
[ <i><tt>Modifiers</tt></i> ]
<i><tt>Type</tt></i> <i><tt>OnType</tt></i>
. <i><tt>Id</tt></i>(<i><tt>Formals</tt></i>)
[ <i><tt>ThrowsClause</tt></i> ]
;
</tt></li></ul></div><p>
The effect of such a declaration is to make <i><tt>OnType</tt></i>
support the new method. Even if <i><tt>OnType</tt></i> is
an interface. Even if the method is neither public nor abstract. So the
following is legal AspectJ code:
</p><pre class="programlisting">
interface Iface {}
aspect A {
private void Iface.m() {
System.err.println("I'm a private method on an interface");
}
void worksOnI(Iface iface) {
// calling a private method on an interface
iface.m();
}
}
</pre><p>
An inter-type constructor declaration looks like
</p><div class="itemizedlist"><ul><li><a name="d0e6459"></a><tt>
[ <i><tt>Modifiers</tt></i> ]
<i><tt>OnType</tt></i> . new (
<i><tt>Formals</tt></i> )
[ <i><tt>ThrowsClause</tt></i> ]
{ <i><tt>Body</tt></i> }</tt></li></ul></div><p>
The effect of such a declaration is to make
<i><tt>OnType</tt></i> support the new constructor. It is
an error for <i><tt>OnType</tt></i> to be an interface.
</p><p>
Inter-type declared constructors cannot be used to assign a
value to a final variable declared in <i><tt>OnType</tt></i>.
This limitation significantly increases the ability to both understand
and compile the <i><tt>OnType</tt></i> class and the
declaring aspect separately.
</p><p>
Note that in the Java language, classes that define no constructors
have an implicit no-argument constructor that just calls
<tt>super()</tt>. This means that attempting to declare
a no-argument inter-type constructor on such a class may result in
a conflict, even though it <span class="emphasis"><i>looks</i></span> like no
constructor is defined.
</p><p>
An inter-type field declaration looks like one of
</p><div class="itemizedlist"><ul><li><a name="d0e6504"></a><tt>
[ <i><tt>Modifiers</tt></i> ]
<i><tt>Type</tt></i>
<i><tt>OnType</tt></i> . <i><tt>Id</tt></i>
= <i><tt>Expression</tt></i>;</tt></li><li><a name="d0e6522"></a><tt>
[ <i><tt>Modifiers</tt></i> ]
<i><tt>Type</tt></i>
<i><tt>OnType</tt></i> . <i><tt>Id</tt></i>;</tt></li></ul></div><p>
The effect of such a declaration is to make
<i><tt>OnType</tt></i> support the new field. Even if
<i><tt>OnType</tt></i> is an interface. Even if the field is
neither public, nor static, nor final.
</p><p>
The initializer, if any, of an inter-type field declaration runs
before the class-local initializers defined in its target class.
</p></div><p>
Any occurrence of the identifier <tt>this</tt> in the body of
an inter-type constructor or method declaration, or in the initializer
of an inter-type field declaration, refers to the
<i><tt>OnType</tt></i> object rather than to the aspect
type; it is an error to access <tt>this</tt> in such a
position from a <tt>static</tt> inter-type member
declaration.
</p><div class="sect2"><a name="access-modifiers"></a><div class="titlepage"><div><h3 class="title"><a name="access-modifiers"></a>Access modifiers</h3></div></div><p>
Inter-type member declarations may be public or private, or have
default (package-protected) visibility. AspectJ does not provide
protected inter-type members.
</p><p>
The access modifier applies in relation to the aspect, not in relation
to the target type. So a private inter-type member is visible only from
code that is defined within the declaring aspect. A default-visibility
inter-type member is visible only from code that is defined within the
declaring aspect's package.
</p><p>
Note that a declaring a private inter-type method (which AspectJ
supports) is very different from inserting a private method declaration
into another class. The former allows access only from the declaring
aspect, while the latter would allow access only from the target type.
Java serialization, for example, uses the presense of a private method
<tt>void writeObject(ObjectOutputStream)</tt> for the
implementation of <tt>java.io.Serializable</tt>. A private
inter-type declaration of that method would not fulfill this
requirement, since it would be private to the aspect, not private to
the target type.
</p><p>
The access modifier of abstract inter-type methods has
one constraint: It is illegal to declare an abstract
non-public inter-type method on a public interface. This
is illegal because it would say that a public interface
has a constraint that only non-public implementors must
fulfill. This would not be compatible with Java's type
system.
</p></div><div class="sect2"><a name="conflicts"></a><div class="titlepage"><div><h3 class="title"><a name="conflicts"></a>Conflicts</h3></div></div><p>
Inter-type declarations raise the possibility of conflicts among
locally declared members and inter-type members. For example, assuming
<tt>otherPackage</tt> is not the package containing the
aspect <tt>A</tt>, the code
</p><pre class="programlisting">
aspect A {
private Registry otherPackage.onType.r;
public void otherPackage.onType.register(Registry r) {
r.register(this);
this.r = r;
}
}
</pre><p>
declares that <tt>onType</tt> in <tt>otherPackage</tt> has a field
<tt>r</tt>. This field, however, is only accessible from the
code inside of aspect <tt>A</tt>. The aspect also declares
that <tt>onType</tt> has a method
"<tt>register</tt>", but makes this method accessible from
everywhere.
</p><p>
If <tt>onType</tt> already defines a
private or package-protected field "<tt>r</tt>", there is no
conflict: The aspect cannot see such a field, and no code in
<tt>otherPackage</tt> can see the inter-type
"<tt>r</tt>".
</p><p>
If <tt>onType</tt> defines a public field
"<tt>r</tt>", there is a conflict: The expression
</p><pre class="programlisting">
this.r = r
</pre><p>
is an error, since it is ambiguous whether the private inter-type
"<tt>r</tt>" or the public locally-defined
"<tt>r</tt>" should be used.
</p><p>
If <tt>onType</tt> defines a method
"<tt>register(Registry)</tt>" there is a conflict, since it
would be ambiguous to any code that could see such a defined method
which "<tt>register(Registry)</tt>" method was applicable.
</p><p>
Conflicts are resolved as much as possible as per Java's conflict
resolution rules:
</p><div class="itemizedlist"><ul><li><a name="d0e6657"></a>A subclass can inherit multiple <span class="emphasis"><i>fields</i></span> from its superclasses,
all with the same name and type. However, it is an error to have an ambiguous
<span class="emphasis"><i>reference</i></span> to a field.</li><li><a name="d0e6665"></a>A subclass can only inherit multiple
<span class="emphasis"><i>methods</i></span> with the same name and argument types from
its superclasses if only zero or one of them is concrete (i.e., all but
one is abstract, or all are abstract).
</li></ul></div><p>
Given a potential conflict between inter-type member declarations in
different aspects, if one aspect has precedence over the other its
declaration will take effect without any conflict notice from compiler.
This is true both when the precedence is declared explicitly with
<tt>declare precedence</tt> as well as when when sub-aspects
implicitly have precedence over their super-aspect.
</p></div><div class="sect2"><a name="extension-and-implementation"></a><div class="titlepage"><div><h3 class="title"><a name="extension-and-implementation"></a>Extension and Implementation</h3></div></div><p>
An aspect may change the inheritance hierarchy of a system by changing
the superclass of a type or adding a superinterface onto a type, with
the <tt>declare parents</tt> form.
</p><div class="itemizedlist"><ul><li><a name="d0e6684"></a><tt>declare parents: <i><tt>TypePattern</tt></i> extends <i><tt>Type</tt></i>;</tt></li><li><a name="d0e6693"></a><tt>declare parents: <i><tt>TypePattern</tt></i> implements <i><tt>TypeList</tt></i>;</tt></li></ul></div><p>
For example, if an aspect wished to make a particular class runnable,
it might define appropriate inter-type <tt>void
run()</tt> method, but it should also declare that the class
fulfills the <tt>Runnable</tt> interface. In order to
implement the methods in the <tt>Runnable</tt> interface, the
inter-type <tt>run()</tt> method must be public:
</p><pre class="programlisting">
aspect A {
declare parents: SomeClass implements Runnable;
public void SomeClass.run() { ... }
}
</pre></div><div class="sect2"><a name="interfaces-with-members"></a><div class="titlepage"><div><h3 class="title"><a name="interfaces-with-members"></a>Interfaces with members</h3></div></div><p>
Through the use of inter-type members, interfaces may now carry
(non-public-static-final) fields and (non-public-abstract) methods that
classes can inherit. Conflicts may occur from ambiguously inheriting
members from a superclass and multiple superinterfaces.
</p><p>
Because interfaces may carry non-static initializers, each interface
behaves as if it has a zero-argument constructor containing its
initializers. The order of super-interface instantiation is
observable. We fix this order with the following properties: A
supertype is initialized before a subtype, initialized code runs only
once, and the initializers for a type's superclass are run before the
initializers for its superinterfaces. Consider the following hierarchy
where {<tt>Object</tt>, <tt>C</tt>,
<tt>D</tt>, <tt>E</tt>} are classes,
{<tt>M</tt>, <tt>N</tt>, <tt>O</tt>,
<tt>P</tt>, <tt>Q</tt>} are interfaces.
</p><pre class="programlisting">
Object M O
\ / \ /
C N Q
\ / /
D P
\ /
E
</pre><p>
when a new <tt>E</tt> is instantiated, the initializers run in this order:
</p><pre class="programlisting">
Object M C O N D Q P E
</pre></div><div class="sect2"><a name="warnings-and-errors"></a><div class="titlepage"><div><h3 class="title"><a name="warnings-and-errors"></a>Warnings and Errors</h3></div></div><p>An aspect may specify that a particular join point should never be
reached. </p><div class="itemizedlist"><ul><li><a name="d0e6768"></a><tt>declare error: <i><tt>Pointcut</tt></i>: <i><tt>String</tt></i>;</tt></li><li><a name="d0e6777"></a><tt>declare warning: <i><tt>Pointcut</tt></i>: <i><tt>String</tt></i>;</tt></li></ul></div><p>If the compiler determines that a join point in
<i><tt>Pointcut</tt></i> could possibly be reached, then it
will signal either an error or warning, as declared, using the
<i><tt>String</tt></i> for its message. </p></div><div class="sect2"><a name="softened-exceptions"></a><div class="titlepage"><div><h3 class="title"><a name="softened-exceptions"></a>Softened exceptions</h3></div></div><p>An aspect may specify that a particular kind of exception, if
thrown at a join point, should bypass Java's usual static exception
checking system and instead be thrown as a
<tt>org.aspectj.lang.SoftException</tt>, which is subtype of
<tt>RuntimeException</tt> and thus does not need to be
declared. </p><div class="itemizedlist"><ul><li><a name="d0e6806"></a><tt>declare soft: <i><tt>Type</tt></i>: <i><tt>Pointcut</tt></i>;</tt></li></ul></div><p>For example, the aspect</p><pre class="programlisting">
aspect A {
declare soft: Exception: execution(void main(String[] args));
}
</pre><p>Would, at the execution join point, catch any
<tt>Exception</tt> and rethrow a
<tt>org.aspectj.lang.SoftException</tt> containing
original exception. </p><p>This is similar to what the following advice would do</p><pre class="programlisting">
aspect A {
void around() execution(void main(String[] args)) {
try { proceed(); }
catch (Exception e) {
throw new org.aspectj.lang.SoftException(e);
}
}
}
</pre><p>except, in addition to wrapping the exception, it also affects
Java's static exception checking mechanism. </p><p> Like advice, the declare soft form has no effect in an
abstract aspect that is not extended by a concreate aspect. So
the following code will not compile unless it is compiled with an
extending concrete aspect:</p><pre class="programlisting">
abstract aspect A {
abstract pointcut softeningPC();
before() : softeningPC() {
Class.forName("FooClass"); // error: uncaught ClassNotFoundException
}
declare soft : ClassNotFoundException : call(* Class.*(..));
}
</pre></div><div class="sect2"><a name="advice-precedence"></a><div class="titlepage"><div><h3 class="title"><a name="advice-precedence"></a>Advice Precedence</h3></div></div><p>
An aspect may declare a precedence relationship between concrete
aspects with the <tt>declare precedence</tt> form:
</p><div class="itemizedlist"><ul><li><a name="d0e6846"></a><tt>declare precedence :
<i><tt>TypePatternList</tt></i> ; </tt></li></ul></div><p>This signifies that if any join point has advice from two
concrete aspects matched by some pattern in
<i><tt>TypePatternList</tt></i>, then the precedence of
the advice will be the order of in the list. </p><p>In <i><tt>TypePatternList</tt></i>, the wildcard "*" can
appear at most once, and it means "any type not matched by any other
pattern in the list". </p><p>For example, the constraints that (1) aspects that have
Security as part of their name should have precedence over all other
aspects, and (2) the Logging aspect (and any aspect that extends it)
should have precedence over all non-security aspects, can be
expressed by:</p><pre class="programlisting">
declare precedence: *..*Security*, Logging+, *;
</pre><p>
For another example, the CountEntry aspect might want to count the
entry to methods in the current package accepting a Type object as
its first argument. However, it should count all entries, even
those that the aspect DisallowNulls causes to throw exceptions.
This can be accomplished by stating that CountEntry has precedence
over DisallowNulls. This declaration could be in either aspect, or
in another, ordering aspect:
</p><pre class="programlisting">
aspect Ordering {
declare precedence: CountEntry, DisallowNulls;
}
aspect DisallowNulls {
pointcut allTypeMethods(Type obj): call(* *(..)) &amp;&amp; args(obj, ..);
before(Type obj): allTypeMethods(obj) {
if (obj == null) throw new RuntimeException();
}
}
aspect CountEntry {
pointcut allTypeMethods(Type obj): call(* *(..)) &amp;&amp; args(obj, ..);
static int count = 0;
before(): allTypeMethods(Type) {
count++;
}
}
</pre><div class="sect3"><a name="d0e6870"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6870"></a>Various cycles</h4></div></div><p>
It is an error for any aspect to be matched by more than one
TypePattern in a single decare precedence, so:
</p><pre class="programlisting">
declare precedence: A, B, A ; // error
</pre><p>
However, multiple declare precedence forms may legally have this
kind of circularity. For example, each of these declare
precedence is perfectly legal:
</p><pre class="programlisting">
declare precedence: B, A;
declare precedence: A, B;
</pre><p>
And a system in which both constraints are active may also be
legal, so long as advice from A and B don't share a join
point. So this is an idiom that can be used to enforce that A and
B are strongly independent.
</p></div><div class="sect3"><a name="d0e6883"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6883"></a>Applies to concrete aspects</h4></div></div><p>
Consider the following library aspects:
</p><pre class="programlisting">
abstract aspect Logging {
abstract pointcut logged();
before(): logged() {
System.err.println("thisJoinPoint: " + thisJoinPoint);
}
}
abstract aspect MyProfiling {
abstract pointcut profiled();
Object around(): profiled() {
long beforeTime = System.currentTimeMillis();
try {
return proceed();
} finally {
long afterTime = System.currentTimeMillis();
addToProfile(thisJoinPointStaticPart,
afterTime - beforeTime);
}
}
abstract void addToProfile(
org.aspectj.JoinPoint.StaticPart jp,
long elapsed);
}
</pre><p>
In order to use either aspect, they must be extended with
concrete aspects, say, MyLogging and MyProfiling. Because advice
only applies from concrete aspects, the declare precedence form
only matters when declaring precedence with concrete aspects. So
</p><pre class="programlisting">
declare precedence: Logging, Profiling;
</pre><p>
has no effect, but both
</p><pre class="programlisting">
declare precedence: MyLogging, MyProfiling;
declare precedence: Logging+, Profiling+;
</pre><p>
are meaningful.
</p></div></div><div class="sect2"><a name="statically-determinable-pointcuts"></a><div class="titlepage"><div><h3 class="title"><a name="statically-determinable-pointcuts"></a>Statically determinable pointcuts</h3></div></div><p>Pointcuts that appear inside of <tt>declare</tt> forms
have certain restrictions. Like other pointcuts, these pick out join
points, but they do so in a way that is statically determinable. </p><p>Consequently, such pointcuts may not include, directly or
indirectly (through user-defined pointcut declarations) pointcuts that
discriminate based on dynamic (runtime) context. Therefore, such
pointcuts may not be defined in terms of</p><div class="itemizedlist"><ul><li><a name="d0e6911"></a>cflow</li><li><a name="d0e6913"></a>cflowbelow</li><li><a name="d0e6915"></a>this</li><li><a name="d0e6917"></a>target</li><li><a name="d0e6919"></a>args</li><li><a name="d0e6921"></a>if</li></ul></div><p> all of which can discriminate on runtime information. </p></div></div><div class="sect1"><a name="semantics-aspects"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="semantics-aspects"></a>Aspects</h2></div></div><p>
An aspect is a crosscutting type defined by the <tt>aspect</tt>
declaration.
</p><div class="sect2"><a name="aspect-declaration"></a><div class="titlepage"><div><h3 class="title"><a name="aspect-declaration"></a>Aspect Declaration</h3></div></div><p>
The <tt>aspect</tt> declaration is similar to the
<tt>class</tt> declaration in that it defines a type and an
implementation for that type. It differs in a number of
ways:
</p><div class="sect3"><a name="d0e6944"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6944"></a>Aspect implementation can cut across other types</h4></div></div><p> In addition to normal Java class declarations such as
methods and fields, aspect declarations can include AspectJ
declarations such as advice, pointcuts, and inter-type
declarations. Thus, aspects contain implementation
declarations that can can cut across other types (including those defined by
other aspect declarations).
</p></div><div class="sect3"><a name="d0e6949"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6949"></a>Aspects are not directly instantiated</h4></div></div><p> Aspects are not directly instantiated with a new
expression, with cloning, or with serialization. Aspects may
have one constructor definition, but if so it must be of a
constructor taking no arguments and throwing no checked
exceptions.
</p></div><div class="sect3"><a name="d0e6954"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6954"></a>Nested aspects must be <tt>static</tt></h4></div></div><p>
Aspects may be defined either at the package level, or as a static nested
aspect -- that is, a static member of a class, interface, or aspect. If it
is not at the package level, the aspect <span class="emphasis"><i>must</i></span> be
defined with the static keyword. Local and anonymous aspects are not
allowed.
</p></div></div><div class="sect2"><a name="aspect-extension"></a><div class="titlepage"><div><h3 class="title"><a name="aspect-extension"></a>Aspect Extension</h3></div></div><p>
To support abstraction and composition of crosscutting concerns,
aspects can be extended in much the same way that classes can. Aspect
extension adds some new rules, though.
</p><div class="sect3"><a name="d0e6969"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6969"></a>Aspects may extend classes and implement interfaces</h4></div></div><p>
An aspect, abstract or concrete, may extend a class and may implement
a set of interfaces. Extending a class does not provide the ability
to instantiate the aspect with a new expression: The aspect may still
only define a null constructor.
</p></div><div class="sect3"><a name="d0e6974"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6974"></a>Classes may not extend aspects</h4></div></div><p>
It is an error for a class to extend or implement an aspect.
</p></div><div class="sect3"><a name="d0e6979"></a><div class="titlepage"><div><h4 class="title"><a name="d0e6979"></a>Aspects extending aspects
</h4></div></div><p>
Aspects may extend other aspects, in which case not only are fields
and methods inherited but so are pointcuts. However, aspects may only
extend abstract aspects. It is an error for a concrete aspect to
extend another concrete aspect.
</p></div></div><div class="sect2"><a name="aspect-instantiation"></a><div class="titlepage"><div><h3 class="title"><a name="aspect-instantiation"></a>Aspect instantiation</h3></div></div><p>
Unlike class expressions, aspects are not instantiated with
<tt>new</tt> expressions. Rather, aspect instances are
automatically created to cut across programs. A program
can get a reference to an aspect instance using the static
method <tt>aspectOf(..)</tt>.
</p><p>
Because advice only runs in the context of an aspect instance, aspect
instantiation indirectly controls when advice runs.
</p><p>
The criteria used to determine how an aspect is instantiated
is inherited from its parent aspect. If the aspect has no parent
aspect, then by default the aspect is a singleton aspect.
How an aspect is instantiated controls the form of the
<tt>aspectOf(..)</tt> method defined on the
concrete aspect class.
</p><div class="sect3"><a name="d0e7002"></a><div class="titlepage"><div><h4 class="title"><a name="d0e7002"></a>Singleton Aspects</h4></div></div><div class="itemizedlist"><ul><li><a name="d0e7006"></a><tt>aspect <i><tt>Id</tt></i> { ... }</tt></li><li><a name="d0e7012"></a><tt>aspect <i><tt>Id</tt></i> issingleton() { ... }</tt></li></ul></div><p>
By default (or by using the modifier <tt>issingleton()</tt>)
an aspect has exactly one instance that cuts across the entire
program. That instance is available at any time during program
execution from the static method <tt>aspectOf()</tt>
automatically defined on all concrete aspects
-- so, in the above examples, <tt>A.aspectOf()</tt> will
return A's instance. This aspect instance is created as the aspect's
classfile is loaded.
</p><p>
Because the an instance of the aspect exists at all join points in
the running of a program (once its class is loaded), its advice will
have a chance to run at all such join points.
</p><p>
(In actuality, one instance of the aspect A is made for each version
of the aspect A, so there will be one instantiation for each time A
is loaded by a different classloader.)
</p></div><div class="sect3"><a name="d0e7033"></a><div class="titlepage"><div><h4 class="title"><a name="d0e7033"></a>Per-object aspects</h4></div></div><div class="itemizedlist"><ul><li><a name="d0e7037"></a><tt>aspect <i><tt>Id</tt></i> perthis(<i><tt>Pointcut</tt></i>) { ... }</tt></li><li><a name="d0e7046"></a><tt>aspect <i><tt>Id</tt></i> pertarget(<i><tt>Pointcut</tt></i>) { ... }</tt></li></ul></div><p>
If an aspect A is defined
<tt>perthis(<i><tt>Pointcut</tt></i>)</tt>, then
one object of type A is created for every object that is the
executing object (i.e., "this") at any of the join points picked out
by <i><tt>Pointcut</tt></i>.
The advice defined in A will run only at a join point where the
currently executing object has been associated with an instance of
A.
</p><p> Similarly, if an aspect A is defined
<tt>pertarget(<i><tt>Pointcut</tt></i>)</tt>,
then one object of type A is created for every object that is the
target object of the join points picked out by
<i><tt>Pointcut</tt></i>.
The advice defined in A will run only at a join point where the
target object has been associated with an instance of
A.
</p><p>
In either case, the static method call
<tt>A.aspectOf(Object)</tt> can be used to get the aspect
instance (of type A) registered with the object. Each aspect
instance is created as early as possible, but not before reaching a
join point picked out by <i><tt>Pointcut</tt></i> where
there is no associated aspect of type A.
</p><p> Both <tt>perthis</tt> and <tt>pertarget</tt>
aspects may be affected by code the AspectJ compiler controls, as
discussed in the <a href="#implementation">Implementation Notes</a> appendix. </p></div><div class="sect3"><a name="d0e7095"></a><div class="titlepage"><div><h4 class="title"><a name="d0e7095"></a>Per-control-flow aspects</h4></div></div><div class="itemizedlist"><ul><li><a name="d0e7099"></a><tt>aspect <i><tt>Id</tt></i> percflow(<i><tt>Pointcut</tt></i>) { ... }</tt></li><li><a name="d0e7108"></a><tt>aspect <i><tt>Id</tt></i> percflowbelow(<i><tt>Pointcut</tt></i>) { ... }</tt></li></ul></div><p>
If an aspect A is defined
<tt>percflow(<i><tt>Pointcut</tt></i>)</tt> or
<tt>percflowbelow(<i><tt>Pointcut</tt></i>)</tt>,
then one object of type A is created for each flow of control of the
join points picked out by <i><tt>Pointcut</tt></i>, either
as the flow of control is entered, or below the flow of control,
respectively. The advice defined in A may run at any join point in
or under that control flow. During each such flow of control, the
static method <tt>A.aspectOf()</tt> will return an object
of type
A. An instance of the aspect is created upon entry into each such
control flow.
</p></div><div class="sect3"><a name="d0e7137"></a><div class="titlepage"><div><h4 class="title"><a name="d0e7137"></a>Aspect instantiation and advice</h4></div></div><p>
All advice runs in the context of an aspect instance,
but it is possible to write a piece of advice with a pointcut
that picks out a join point that must occur before asopect
instantiation. For example:
</p><pre class="programlisting">
public class Client
{
public static void main(String[] args) {
Client c = new Client();
}
}
aspect Watchcall {
pointcut myConstructor(): execution(new(..));
before(): myConstructor() {
System.err.println("Entering Constructor");
}
}
</pre><p>
The before advice should run before the execution of all
constructors in the system. It must run in the context of an
instance of the Watchcall aspect. The only way to get such an
instance is to have Watchcall's default constructor execute. But
before that executes, we need to run the before advice...
</p><p>
There is no general way to detect these kinds of circularities at
compile time. If advice runs before its aspect is instantiated,
AspectJ will throw a <a href="../api/org/aspectj/lang/NoAspectBoundException.html" target="_top">
<tt>org.aspectj.lang.NoAspectBoundException</tt></a>.
</p></div></div><div class="sect2"><a name="aspect-privilege"></a><div class="titlepage"><div><h3 class="title"><a name="aspect-privilege"></a>Aspect privilege</h3></div></div><div class="itemizedlist"><ul><li><a name="d0e7157"></a><tt>privileged aspect <i><tt>Id</tt></i> { ... }</tt></li></ul></div><p>
Code written in aspects is subject to the same access control rules as
Java code when referring to members of classes or aspects. So, for
example, code written in an aspect may not refer to members with
default (package-protected) visibility unless the aspect is defined in
the same package.
</p><p>
While these restrictions are suitable for many aspects, there may be
some aspects in which advice or inter-type members needs to access private
or protected resources of other types. To allow this, aspects may be
declared <tt>privileged</tt>. Code in priviliged aspects has
access to all members, even private ones.
</p><pre class="programlisting">
class C {
private int i = 0;
void incI(int x) { i = i+x; }
}
privileged aspect A {
static final int MAX = 1000;
before(int x, C c): call(void C.incI(int)) &amp;&amp; target(c) &amp;&amp; args(x) {
if (c.i+x &gt; MAX) throw new RuntimeException();
}
}
</pre><p>
In this case, if A had not been declared privileged, the field reference
c.i would have resulted in an error signaled by the compiler.
</p><p>
If a privileged aspect can access multiple versions of a particular
member, then those that it could see if it were not privileged take
precedence. For example, in the code
</p><pre class="programlisting">
class C {
private int i = 0;
void foo() { }
}
privileged aspect A {
private int C.i = 999;
before(C c): call(void C.foo()) target(c) {
System.out.println(c.i);
}
}
</pre><p>
A's private inter-type field C.i, initially bound to 999, will be
referenced in the body of the advice in preference to C's privately
declared field, since the A would have access to its own inter-type
fields even if it were not privileged.
</p><p>
Note that a privileged aspect can access private inter-type
declarations made by other aspects, since they are simply
considered private members of that other aspect.
</p></div></div></div><div class="appendix"><div class="titlepage"><div><h2 class="title"><a name="implementation"></a>Appendix C. Implementation Notes</h2></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><a href="#d0e7185">Compiler Notes</a></dt><dt><a href="#d0e7259">Bytecode Notes</a></dt><dd><dl><dt><a href="#the-class-expression-and-string">The .class expression and String +</a></dt><dt><a href="#the-handler-join-point">The Handler join point</a></dt><dt><a href="#initializers-and-inter-type-constructors">Initializers and Inter-type Constructors</a></dt></dl></dd><dt><a href="#d0e7365">Summary of implementation requirements</a></dt></dl></div><div class="sect1"><a name="d0e7185"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="d0e7185"></a>Compiler Notes</h2></div></div><p>
The initial implementations of AspectJ have all been
compiler-based implementations. Certain elements of AspectJ's
semantics are difficult to implement without making modifications
to the virtual machine, which a compiler-based implementation
cannot do. One way to deal with this problem would be to specify
only the behavior that is easiest to implement. We have chosen a
somewhat different approach, which is to specify an ideal language
semantics, as well as a clearly defined way in which
implementations are allowed to deviate from that semantics. This
makes it possible to develop conforming AspectJ implementations
today, while still making it clear what later, and presumably
better, implementations should do tomorrow.
</p><p>
According to the AspectJ language semantics, the declaration
</p><pre class="programlisting">
before(): get(int Point.x) { System.out.println("got x"); }
</pre><p>
should advise all accesses of a field of type int and name x from
instances of type (or subtype of) Point. It should do this
regardless of whether all the source code performing the access
was available at the time the aspect containing this advice was
compiled, whether changes were made later, etc.
</p><p>
But AspectJ implementations are permitted to deviate from this in
a well-defined way -- they are permitted to advise only accesses
in <span class="emphasis"><i>code the implementation controls</i></span>. Each
implementation is free within certain bounds to provide its own
definition of what it means to control code.
</p><p>
In the current AspectJ compiler, ajc, control of the code means
having bytecode for any aspects and all the code they should
affect available during the compile. This means that if some class
Client contains code with the expression <tt>new
Point().x</tt> (which results in a field get join point at
runtime), the current AspectJ compiler will fail to advise that
access unless Client.java or Client.class is compiled as well. It
also means that join points associated with code in native methods
(including their execution join points) cannot be advised.
</p><p>
Different join points have different requirements. Method and
constructor call join points can be advised only if ajc controls
the bytecode for the caller. Field reference or assignment join
points can be advised only if ajc controls the bytecode for the
"caller", the code actually making the reference or assignment.
Initialization join points can be advised only if ajc controls the
bytecode of the type being initialized, and execution join points
can be advised only if ajc controls the bytecode for the method or
constructor body in question.
The end of an exception handler is underdetermined in bytecode,
so ajc will not implement after or around advice on handler join
points.
Similarly, ajc cannot implement around advice on initialization
or preinitialization join points.
In cases where ajc cannot implement advice, it will emit a
compile-time error noting this as a compiler limitation.
</p><p>
Aspects that are defined <tt>perthis</tt> or
<tt>pertarget</tt> also have restrictions based on
control of the code. In particular, at a join point where the
bytecode for the currently executing object is not available, an
aspect defined <tt>perthis</tt> of that join point will
not be associated. So aspects defined
<tt>perthis(Object)</tt> will not create aspect
instances for every object unless <tt>Object</tt>is part
of the compile. Similar restrictions apply to
<tt>pertarget</tt> aspects.
</p><p>
Inter-type declarations such as <tt>declare parents</tt>
also have restrictions based on control of the code. If the
bytecode for the target of an inter-type declaration is not
available, then the inter-type declaration is not made on that
target. So, <tt>declare parents : String implements
MyInterface</tt> will not work for
<tt>java.lang.String</tt> unless
<tt>java.lang.String</tt> is part of the compile.
</p><p>
When declaring members on interfaces, the implementation must
control both the interface and the top-level implementors of
that interface (the classes that implement the interface but
do not have a superclass that implements the interface).
You may weave these separately, but be aware that you will get
runtime exceptions if you run the affected top-level classes
without the interface as produced by the same ajc implementation.
Any intertype declaration of an abstract method on an interface
must be specified as public, you will get a compile time error
message indicating this is a compiler limitation if you do not
specify public. A non-abstract method declared on an interface
can use any access modifier except protected. Note that this is
different to normal Java rules where all members declared in
an interface are implicitly public.
Finally, note that one cannot define static fields or methods
on interfaces.
</p><p>
When declaring methods on target types, only methods declared
public are recognizable in the bytecode, so methods must be
declared public to be overridden in any subtype or to be called
from code in a later compile using the target type as a library.
</p><p>
Other AspectJ implementations, indeed, future versions of ajc, may
define <span class="emphasis"><i>code the implementation controls</i></span> more
liberally or restrictively, so long as they comport with the Java
language. For example, the <tt>call</tt> pointcut does
not pick out reflective calls to a method implemented in
<tt>java.lang.reflect.Method.invoke(Object, Object[])</tt>.
Some suggest that the call "happens" and the call pointcut should
pick it out, but the AspectJ language shouldn't anticipate what happens
in code outside the control of the implementation, even when it
is a a well-defined API in a Java standard library.
</p><p>
The important thing to remember is that core concepts of AspectJ,
such as the join point, are unchanged, regardless of which
implementation is used. During your development, you will have to
be aware of the limitations of the ajc compiler you're using, but
these limitations should not drive the design of your aspects.
</p></div><div class="sect1"><a name="d0e7259"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="d0e7259"></a>Bytecode Notes</h2></div></div><div class="sect2"><a name="the-class-expression-and-string"></a><div class="titlepage"><div><h3 class="title"><a name="the-class-expression-and-string"></a>The .class expression and String +</h3></div></div><p> The java language form <tt>Foo.class</tt> is
implemented in bytecode with a call to
<tt>Class.forName</tt> guarded by an exception
handler catching a <tt>ClassNotFoundException</tt>.
</p><p> The java language + operator, when applied to String
arguments, is implemented in bytecode by calls to
<tt>StringBuffer.append</tt>.
</p><p> In both of these cases, the current AspectJ compiler
operates on the bytecode implementation of these language
features; in short, it operates on what is really happening rather
than what was written in source code. This means that there may
be call join points to <tt>Class.forName</tt> or
<tt>StringBuffer.append</tt> from programs that do not,
at first glance, appear to contain such calls:
</p><pre class="programlisting">
class Test {
void main(String[] args) {
System.out.println(Test.class); // calls Class.forName
System.out.println(args[0] + args[1]); // calls StringBuffer.append
}
}
</pre><p>In short, the join point model of the current AspectJ
compiler considers these as valid join points.
</p></div><div class="sect2"><a name="the-handler-join-point"></a><div class="titlepage"><div><h3 class="title"><a name="the-handler-join-point"></a>The Handler join point</h3></div></div><p>The end of exception handlers cannot reliably be found in Java
bytecode. Instead of removing the handler join point entirely, the
current AspectJ compiler restricts what can be done with the handler
join point:
</p><div class="itemizedlist"><ul><li><a name="d0e7299"></a>After and around advice cannot apply to handler
join points.</li><li><a name="d0e7301"></a>The control flow of a handler join point cannot be
detected. </li></ul></div><p>
The first of these is relatively straightforward. If any piece of
after advice (returning, throwing, or "finally") would normally
apply to a handler join point, it will not in code output by the
current AspectJ compiler. A compiler warning is generated whenever
this is detected to be the case. Before advice is allowed.
</p><p> The second is that the control flow of a handler join point
is not picked out. For example, the following pointcut
</p><pre class="programlisting">
cflow(call(void foo()) || handler(java.io.IOException))
</pre><p> will capture all join points in the control flow of a call to
<tt>void foo()</tt>, but it will <span class="emphasis"><i>not</i></span>
capture those in the control flow of an
<tt>IOException</tt> handler. It is equivalent to
<tt>cflow(call(void foo()))</tt>. In general,
<tt>cflow(handler(<i><tt>Type</tt></i>))</tt>
will not pick out any join points, the one exception to this is join points
that occur during the execution of any before advice on the handler.
</p><p> This does not restrict programs from placing before advice on
handlers inside <span class="emphasis"><i>other</i></span> control flows. This
advice, for example, is perfectly fine:
</p><pre class="programlisting">
before(): handler(java.io.IOException) &amp;&amp; cflow(void parse()) {
System.out.println("about to handle an exception while parsing");
}
</pre><p>
A source-code implementation of AspectJ (such as AspectJ 1.0.6) is
able to detect the endpoint of a handler join point, and as such
will likely have fewer such restrictions.
</p></div><div class="sect2"><a name="initializers-and-inter-type-constructors"></a><div class="titlepage"><div><h3 class="title"><a name="initializers-and-inter-type-constructors"></a>Initializers and Inter-type Constructors</h3></div></div><p>
The code for Java initializers, such as the assignment to the
field d in
</p><pre class="programlisting">
class C {
double d = Math.sqrt(2);
}
</pre><p>
are considered part of constructors by the time AspectJ gets ahold
of bytecode. That is, the assignment of d to the square root of
two happens <span class="emphasis"><i>inside</i></span> the default constructor of
C.
</p><p>
Thus inter-type constructors will not necessarily run a target
type's initialization code. In particular, if the inter-type
constructor calls a super-constructor (as opposed to a
<tt>this</tt> constructor), the target type's
initialization code will <span class="emphasis"><i>not</i></span> be run when that
inter-type constructor is called.
</p><pre class="programlisting">
aspect A {
C.new(Object o) {} // implicitly calls super()
public static void main(String[] args) {
System.out.println((new C() ).d); // prints 1.414...
System.out.println((new C(null)).d); // prints 0.0
}
</pre><p>
It is the job of an inter-type constructor to do all the required
initialization, or to delegate to a <tt>this</tt>
constructor if necessary.
</p></div></div><div class="sect1"><a name="d0e7365"></a><div class="titlepage"><div><h2 class="title" style="clear: both"><a name="d0e7365"></a>Summary of implementation requirements</h2></div></div><p>
This summarizes the requirements of our implementation of AspectJ.
For more details, see the relevant sections of this guide.
</p><div class="itemizedlist"><ul compact><li><p><a name="d0e7371"></a>The invoking code must be under the control of ajc
for the following join points:</p><div class="itemizedlist"><ul compact><li><a name="d0e7375"></a>call join point</li><li><a name="d0e7377"></a>get join point</li><li><a name="d0e7379"></a>set join point</li></ul></div></li><li><p><a name="d0e7381"></a>The declaring/target code must be under the control of ajc
for the following join points and inter-type declarations:</p><div class="itemizedlist"><ul compact><li><a name="d0e7385"></a>execution join point</li><li><a name="d0e7387"></a>adviceexecution join point</li><li><a name="d0e7389"></a>handler join point</li><li><a name="d0e7391"></a>initialization join point</li><li><a name="d0e7393"></a>preinitialiaztion join point</li><li><a name="d0e7395"></a>staticinitialization join point</li><li><a name="d0e7397"></a>perthis aspect</li><li><a name="d0e7399"></a>pertarget aspect</li><li><a name="d0e7401"></a>declare parents</li><li><a name="d0e7403"></a>declare method or field (see interface caveats below)</li></ul></div></li><li><p><a name="d0e7405"></a>Implementation Caveats</p><div class="itemizedlist"><ul compact><li><p><a name="d0e7409"></a>The initialization and preinitialization join points
do not support around advice</p></li><li><p><a name="d0e7412"></a>The handler join point does not support...</p><div class="itemizedlist"><ul compact><li><a name="d0e7416"></a>after advice</li><li><a name="d0e7418"></a>around advice</li><li><a name="d0e7420"></a>cflow(handler(..))</li></ul></div></li><li><p><a name="d0e7422"></a>Declaring members on an interface in an aspect affects only
the topmost implementing classes the implementation controls.</p></li><li><p><a name="d0e7425"></a>cflow and cflowbelow pointcuts work within a single thread.</p></li></ul></div></li></ul></div></div></div></div></body></html>