The Build Lifecycle
Chapter 4
4.1. Introduction
Maven models projects as nouns which are described by a POM. The POM captures the identity of a project: What does a project contain? What type of packaging a project needs? Does the project have a parent? What are the dependencies? We’ve explored the idea of describing a project in the previous chapters, but we haven’t introduced the mechanism that allows Maven to act upon these objects. In Maven the "verbs" are goals packaged in Maven plugins which are tied to a phases in a build lifecycle. A Maven lifecycle consists of a sequence of named phases: prepare-resources, compile, package, and install among other. There is phase that captures compilation and a phase that captures packaging. There are pre- and post- phases which can be used to register goals which must run prior to compilation, or tasks which must be run after a particular phase. When you tell Maven to build a project, you are telling Maven to step through a defined sequence of phases and execute any goals which may have been registered with each phase.
A build lifecycle is an organized sequence of phases that exist to give order to a set of goals. Those goals are chosen and bound by the packaging type of the project being acted upon. There are three standard lifecycles in Maven: clean, default (sometimes called build) and site. In this chapter, you are going to learn how Maven ties goals to lifecycle phases and how the lifecycle can be customized. You will also learn about the default lifecycle phases.
4.1.1. Clean Lifecycle (clean)
The first lifecycle you’ll be interested in is the simplest lifecycle in Maven. Running mvn clean
invokes the clean lifecycle which consists of three lifecycle phases:
pre-clean
clean
post-clean
The interesting phase in the clean lifecycle is the clean
phase. The Clean plugin’s clean goal (clean:clean
) is bound to the clean
phase in the clean
lifecycle. The clean:clean
goal deletes the output of a build by deleting the build directory. If you haven’t customized the location of the build directory it will be the ${basedir}/target directory as defined by the Super POM. When you execute the clean:clean
goal you do not do so by executing the goal directly with mvn clean:clean
, you do so by executing the clean
phase of the clean lifecycle. Executing the clean
phase gives Maven an opportunity to execute any other goals which may be bound to the pre-clean
phase.
For example, suppose you wanted to trigger an antrun:run
goal task to echo a notification on pre-clean
, or to make an archive of a project’s build directory before it is deleted. Simply running the clean:clean
goal will not execute the lifecycle at all, but specifying the clean
phase will use the clean
lifecycle and advance through the three lifecycle phases until it reaches the clean
phase. Triggering a Goal on pre-clean shows an example of build configuration which binds the antrun:run
goal to the pre-clean
phase to echo an alert that the project artifact is about to be deleted. In this example, the antrun:run
goal is being used to execute some arbitrary Ant commands to check for an existing project artifact. If the project’s artifact is about to be deleted it will print this to the screen
Triggering a Goal on pre-clean.
<project> ... <build> <plugins>... <plugin> <artifactId>maven-antrun-plugin</artifactId> <executions> <execution> <id>file-exists</id> <phase>pre-clean</phase> <goals> <goal>run</goal> </goals> <configuration> <tasks> <!-- adds the ant-contrib tasks (if/then/else used below) --> <taskdef resource="net/sf/antcontrib/antcontrib.properties" /> <available file="${project.build.directory}/${project.build.finalName}.${project.packaging}" property="file.exists" value="true" /> <if> <not> <isset property="file.exists" /> </not> <then> <echo>No ${project.build.finalName}.${project.packaging} to delete</echo> </then> <else> <echo>Deleting ${project.build.finalName}.${project.packaging}</echo> </else> </if> </tasks> </configuration> </execution> </executions> <dependencies> <dependency> <groupId>ant-contrib</groupId> <artifactId>ant-contrib</artifactId> <version>1.0b2</version> </dependency> </dependencies> </plugin> </plugins> </build> </project>
Running mvn clean
on a project with this build configuration will produce output similar to the following:
[INFO] Scanning for projects... [INFO] ---------------------------------------------------------------------- [INFO] Building Your Project [INFO]task-segment: [clean] [INFO] ---------------------------------------------------------------------- [INFO] [antrun:run {execution: file-exists}] [INFO] Executing tasks [echo] Deleting your-project-1.0-SNAPSHOT.jar [INFO] Executed tasks [INFO] [clean:clean] [INFO] Deleting directory ~/corp/your-project/target [INFO] Deleting directory ~/corp/your-project/target/classes [INFO] Deleting directory ~/corp/your-project/target/test-classes [INFO] ------------------------------------------------------------------------ [INFO] BUILD SUCCESSFUL [INFO] ------------------------------------------------------------------------ [INFO] Total time: 1 second [INFO] Finished at: Wed Nov 08 11:46:26 CST 2006 [INFO] Final Memory: 2M/5M [INFO] ------------------------------------------------------------------------
In addition to configuring Maven to run a goal during the pre-clean
phase, you can also customize the Clean plugin to delete files in addition to the build output directory. You can configure the plugin to remove specific files in a fileSet
. The example below configures clean to remove all .class files in a directory named target-other/ using standard Ant file wildcards: *
and \**
.
Customizing Behavior of the Clean Plugin.
<project> <modelVersion>4.0.0</modelVersion> ... <build> <plugins> <plugin> <artifactId>maven-clean-plugin</artifactId> <configuration> <filesets> <fileset> <directory>target-other</directory> <includes> <include>*.class</include> </includes> </fileset> </filesets> </configuration> </plugin> </plugins> </build> </project>
Most Maven users will be familiar with the default lifecycle. It is a general model of a build process for a software application. The first phase is validate
and the last phase is deploy
. The phases in the default Maven lifecycle are shown in Table 4.1, “Maven Lifecycle Phases”.
Table 4.1. Maven Lifecycle Phases
Lifecycle Phase | Description |
---|---|
validate |
Validate the project is correct and all necessary information is available to complete a build |
generate-sources |
Generate any source code for inclusion in compilation |
process-sources |
Process the source code, for example to filter any values |
generate-resources |
Generate resources for inclusion in the package |
process-resources |
Copy and process the resources into the destination directory, ready for packaging |
compile |
Compile the source code of the project |
process-classes |
Post-process the generated files from compilation, for example to do bytecode enhancement on Java classes |
generate-test-sources |
Generate any test source code for inclusion in compilation |
process-test-sources |
Process the test source code, for example to filter any values |
generate-test-resources |
Create resources for testing |
process-test-resources |
Copy and process the resources into the test destination directory |
test-compile |
Compile the test source code into the test destination directory |
test |
Run tests using a suitable unit testing framework. These tests should not require the code be packaged or deployed |
prepare-package |
Perform any operations necessary to prepare a package before the actual packaging. This often results in an unpacked, processed version of the package (coming in Maven 2.1+) |
package |
Take the compiled code and package it in its distributable format, such as a JAR, WAR, or EAR |
pre-integration-test |
Perform actions required before integration tests are executed. This may involve things such as setting up the required environment |
integration-test |
Process and deploy the package if necessary into an environment where integration tests can be run |
post-integration-test |
Perform actions required after integration tests have been executed. This may include cleaning up the environment |
verify |
Run any checks to verify the package is valid and meets quality criteria |
install |
Install the package into the local repository, for use as a dependency in other projects locally |
deploy |
Copies the final package to the remote repository for sharing with other developers and projects (usually only relevant during a formal release) |
Maven does more than build software artifacts from project, it can also generate project documentation and reports about the project, or a collection of projects. Project documentation and site generation have a dedicated lifecycle which contains four phases:
- pre-site
- site
- post-site
- site-deploy
The default goals bound to the site lifecycle is:
- site - site:site
- site-deploy -site:deploy
The packaging type does not usually alter this lifecycle since packaging types are concerned primarily with artifact creation, not with the type of site generated. The Site plugin kicks off the execution of Doxia document generation and other report generation plugins. You can generate a site from a Maven project by running the following command:
$ mvn site
For more information about Maven Site generation, see Chapter 10, Site Generation.
4.2. Package-specific Lifecycles
- 4.2.1. JAR
- 4.2.2. POM
- 4.2.3. Maven Plugin
- 4.2.4. EJB
- 4.2.5. WAR
- 4.2.6. EAR
- 4.2.7. Other Packaging Types
The specific goals bound to each phase default to a set of goals specific to a project’s packaging. A project with packaging jar
has a different set of default goals from a project with a packaging of war
. The packaging
element affects the steps required to build a project. For an example of how the packaging affects the build, consider two projects: one with pom
packaging and the other with jar
packaging. The project with pom
packaging will run the site:attach-descriptor
goal during the package
phase, and the project with jar
packaging will run the jar:jar
goal instead.
The following sections describe the lifecycle for all built-in packaging types in Maven. Use these sections to find out what default goals are mapped to default lifecycle phases.
JAR is the default packaging type, the most common, and thus the most commonly encountered lifecycle configuration. The default goals for the JAR lifecycle are shown in Table 4.2, “Default Goals for JAR Packaging”.
POM is the simplest packaging type. The artifact that it generates is itself only, rather than a JAR, SAR, or EAR. There is no code to test or compile, and there are no resources the process. The default goals for projects with POM packaging are shown in Table 4.3, “Default Goals for POM Packaging”.
This packaging type is similar to JAR packaging type with three additions: plugin:descriptor
, plugin:addPluginArtifactMetadata
, and plugin:updateRegistry
. These goals generate a descriptor file and perform some modifications to the repository data. The default goals for projects with plugin packaging are shown in Table 4.4, “Default Goals for Plugin Packaging”.
Table 4.4. Default Goals for Plugin Packaging
Lifecycle Phase |
Goal |
generate-resources |
plugin:descriptor |
process-resources |
resources:resources |
compile |
compiler:compile |
process-test-resources |
resources:testResources |
test-compile |
compiler:testCompile |
test |
surefire:test |
package |
jar:jar, plugin:addPluginArtifactMetadata |
install |
install:install, plugin:updateRegistry |
deploy |
deploy:deploy |
EJBs, or Enterprise Java Beans, are a common data access mechanism for model-driven development in Enterprise Java. Maven provides support for EJB 2 and 3. Though you must configure the EJB plugin to specifically package for EJB3, else the plugin defaults to 2.1 and looks for the presence of certain EJB configuration files. The default goals for projects with EJB packaging are shown in Table 4.5, “Default Goals for EJB Packaging”.
The WAR packaging type is similar to the JAR and EJB types. The exception being the package
goal of war:war
. Note that the war:war
goal requires a web.xml configuration in your src/main/webapp/WEB-INF directory. The default goals for projects with WAR packaging are shown in Table 4.6, “Default Goals for WAR Packaging”.
EARs are probably the simplest Java EE constructs, consisting primarily of the deployment descriptor application.xml file, some resources and some modules. The EAR plugin has a goal named generate-application-xml
which generates the application.xml based upon the configuration in the EAR project’s POM. The default goals for projects with EAR packaging are shown in Table 4.7, “Default Goals for EAR Packaging”.
This is not an exhaustive list of every packaging type available for Maven. There are a number of packaging formats available through external projects and plugins: the NAR (native archive) packaging type, the SWF and SWC packaging types for projects that produce Adobe Flash and Flex content, and many others. You can also define a custom packaging type and customize the default lifecycle goals to suit your own project packaging requirements.
To use one of these custom packaging types, you need two things: a plugin which defines the lifecycle for a custom packaging type and a repository which contains this plugin. Some custom packaging types are defined in plugins available from the central Maven repository. Here is an example of a project which references the Israfil Flex plugin and uses a custom packaging type of SWF to produce output from Adobe Flex source.
Custom Packaging Type for Adobe Flex (SWF).
<project> ... <packaging>swf</packaging> ... <build> <plugins> <plugin> <groupId>net.israfil.mojo</groupId> <artifactId>maven-flex2-plugin</artifactId> <version>1.4-SNAPSHOT</version> <extensions>true</extensions> <configuration> <debug>true</debug> <flexHome>${flex.home}</flexHome> <useNetwork>true</useNetwork> <main>org/sonatype/mavenbook/Main.mxml</main> </configuration> </plugin> </plugins> </build> ... </project>
In Section 11.6, “Plugins and the Maven Lifecycle”, we show you how to create your own packaging type with a customized lifecycle. This example should give you an idea of what you’ll need to do to reference a custom packaging type. All you need to do is reference the plugin which supplies the custom packaging type. The Israfil Flex plugin is a third-party Maven plugin hosted at Google Code, for more information about this plugin and how to use Maven to compile Adobe Flex go to http://code.google.com/p/israfil-mojo. This plugin supplies the following lifecycle for the SWF packaging type:
- 4.3.1. Process Resources
- 4.3.2. Compile
- 4.3.3. Process Test Resources
- 4.3.4. Test Compile
- 4.3.5. Test
- 4.3.6. Install
- 4.3.7. Deploy
Many of the packaging lifecycles have similar goals. If you look at the goals bound to the WAR and JAR lifecycles, you’ll see that they differ only in the package
phase. The package
phase of the WAR lifecycle calls war:war
and the package
phase of the JAR lifecycle calls jar:jar
. Most of the lifecycles you will come into contact with share some common lifecycle goals for managing resources, running tests, and compiling source code. In this section, we’ll explore some of these common lifecycle goals in detail.
The process-resources
phase "processes" resources and copies them to the output directory. If you haven’t customized the default directory locations defined in the Super POM, this means that Maven will copy the files from ${basedir}/src/main/resources to ${basedir}/target/classes or the directory defined in ${project.build.outputDirectory}. In addition to copying the resources to the output directory, Maven can also apply a filter to the resources that allows you to replace tokens within resource file. Just like variables are referenced in a POM using ${...} notation, you can reference variables in your project’s resources using the same syntax. Coupled with build profiles, such a facility can be used to produce build artifacts which target different deployment platforms. This is something that is common in environments which need to produce output for development, testing, staging, and production platforms from the same project. For more information about build profiles, see Chapter 5, Build Profiles.
To illustrate resource filtering, assume that you have a project with an XML file in src/main/resources/META-INF/service.xml. You want to externalize some configuration variables to a properties file. In other words, you might want to reference a JDBC URL, username, and password for your database, and you don’t want to put these values directly into the service.xml file. Instead, you would like to use a properties file to capture all of the configuration points for your program. Doing this will allow you to consolidate all configuration into a single properties file and make it easier to change configuration values when you need to target a new deployment environment. First, take a look at the contents of service.xml in src/main/resources/META-INF.
Using Properties in Project Resources.
<service> <!-- This URL was set by project version ${project.version} --> <url>${jdbc.url}</url> <user>${jdbc.username}</user> <password>${jdbc.password}</password> </service>
This XML file uses the same property reference syntax you can use in the POM. In fact, the first variable referenced is the project
variable which is also an implicit variable made available in the POM. The project
variable provides access to POM information. The next three variable references are jdbc.url
, jdbc.username
, and jdbc.password
. These custom variables are defined in a properties file src/main/filters/default.properties.
default.properties in src/main/filters.
jdbc.url=jdbc:hsqldb:mem:mydb jdbc.username=sa jdbc.password=
To configure resource filtering with this default.properties file, we need to specify two things in a project’s POM: a list of properties files in the filters
element of the build configuration, and a flag to Maven that the resources directory is to be filtered. The default Maven behavior is to skip filtering and just copy the resources to the output directory; you’ll need to explicitly configure resource filter, or Maven will skip the step altogether. This default ensures that Maven’s resource filtering feature doesn’t surprise you out of nowhere and clobbering any ${...} references you didn’t want it to replace.
Filter Resources (Replacing Properties).
<build> <filters> <filter>src/main/filters/default.properties</filter> </filters> <resources> <resource> <directory>src/main/resources</directory> <filtering>true</filtering> </resource> </resources> </build>
As with all directories in Maven, the resources directory does not need to be in src/main/resources. This is just the default value defined in the Super POM. You should also note that you don’t need to consolidate all of your resources into a single directory. You can always separate resources into separate directories under src/main. Assume that you have a project which contains hundreds of XML documents and hundreds of images. Instead of mixing the resources in the src/main/resources directory, you might want to create two directories src/main/xml and src/main/images to hold this content. To add directories to the list of resource directories, you would add the following resource
elements to your build configuration.
Configuring Additional Resource Directories.
<build> ... <resources> <resource> <directory>src/main/resources</directory> </resource> <resource> <directory>src/main/xml</directory> </resource> <resource> <directory>src/main/images</directory> </resource> </resources> ... </build>
When you are building a project that produces a console application or a command-line tool, you’ll often find yourself writing simple shell scripts that need to reference the JAR produced by a build. When you are using the assembly plugin to produce a distribution for an application as a ZIP or TAR, you might place all of your scripts in a directory like src/main/command. In the following POM resource configuration, you’ll see how we can use resource filtering and a reference to the project variable to capture the final output name of the JAR. For more information about the Maven Assembly plugin, see Chapter 8, Maven Assemblies.
Filtering Script Resources.
<build> <groupId>org.sonatype.mavenbook</groupId> <artifactId>simple-cmd</artifactId> <version>2.3.1</version> ... <resources> <resource> <filtering>true</filtering> <directory>${basedir}/src/main/command</directory> <includes> <include>run.bat</include> <include>run.sh</include> </includes> <targetPath>${basedir}</targetPath> </resource> <resource> <directory>${basedir}/src/main/resources</directory> </resource> </resources> ... </build>
If you run mvn process-resources in this project, you will end up with two files, run.sh and run.bat, in ${basedir}. We’ve singled out these two files in a resource
element, configuring filtering, and set the targetPath
to be ${basedir}. In a second resource
element, we’ve configured the default resources path to be copied to the default output directory without any filtering. Filtering Script Resources shows you how to declare two resource directories and supply them with different filtering and target directory preferences. The project from Filtering Script Resources would contain a run.bat file in src/main/command with the following content:
@echo off java -jar ${project.build.finalName}.jar %*
After running mvn process-resources
, a file named run.bat would appear in ${basedir} with the following content:
@echo off java -jar simple-cmd-2.3.1.jar %*
The ability to customize filtering for specific subsets of resources is another reason why complex projects with many different kinds of resources often find it advantageous to separate resources into multiple directories. The alternative to storing different kinds of resources with different filtering requirements in different directories is to use a more complex set of include and exclude patterns to match all resource files which match a certain pattern.
Most lifecycles bind the Compiler plugin’s compile
goal to the compile
phase. This phase calls out to compile:compile
which is configured to compile all of the source code and copy the bytecode to the build output directory. If you haven’t customized the values defined in the Super POM, compile:compile
is going to compile everything from src/main/java to target/classes. The Compiler plugin calls out to javac and uses default source and target settings of 1.3 and 1.1. In other words, the compiler plugin assumes that your Java source conforms to Java 1.3 and that you are targeting a Java 1.1 JVM. If you would like to change these settings, you’ll need to supply the target and source configuration to the Compiler plugin in your project’s POM as shown in Setting the Source and Target Versions for the Compiler Plugin.
Setting the Source and Target Versions for the Compiler Plugin.
<project> ... <build> ... <plugins> <plugin> <artifactId>maven-compiler-plugin</artifactId> <configuration> <source>1.5</source> <target>1.5</target> </configuration> </plugin> </plugins> ... </build> ... </project>
Notice we are configuring the Compiler plugin, and not the specific compile:compile
goal. If we were going to configure the source and target for just the compile:compile
goal, we would place the configuration
element below an execution
element for the compile:compile
goal. We’ve configured the target and source for the plugin because compile:compile
isn’t the only goal we’re interested in configuring. The Compiler plugin is reused when Maven compiles tests using the compile:testCompile
goal, and configuring target and source at the plugin level allows us to define it once for all goals in a plugin.
If you need to customize the location of the source code, you can do so by changing the build configuration. If you wanted to store your project’s source code in src/java instead of src/main/java and if you wanted build output to go to classes instead of target/classes, you could always override the default sourceDirectory
defined by the Super POM.
Overriding the Default Source Directory.
<build> ... <sourceDirectory>src/java</sourceDirectory> <outputDirectory>classes</outputDirectory> ... </build>
Warning
While it might seem necessary to bend Maven to your own idea of project directory structure, we can’t emphasize enough that you should sacrifice your own ideas of directory structure in favor of the Maven defaults. This isn’t because we’re trying to brainwash you into accepting the Maven Way, but it will be easier for people to understand your project if it adheres to the most basic conventions. Just forget about this. Don’t do it.
The process-test-resources
phase is almost indistinguishable from the process-resources
phase. There are some trivial differences in the POM, but most everything the same. You can filter test resources just as you filter regular resources. The default location for test resources is defined in the Super POM as src/test/resources, and the default output directory for test resources is target/test-classes as defined in ${project.build.testOutputDirectory}.
The test-compile
phase is almost identical to the compile
phase. The only difference is that test-compile
is going to invoke compile:testCompile
to compile source from the test source directory to the test build output directory. If you haven’t customized the default directories from the Super POM, compile:testCompile
is going to compile the source in src/test/java to the target/test-classes directory.
As with the source code directory, if you want to customize the location of the test source code and the output of test compilation, you can do so by overriding the testSourceDirectory
and the testOutputDirectory. If you wanted to store test source in src-test/ instead of src/test/java and you wanted to save test bytecode to classes-test/ instead of target/test-classes, you would use the following configuration.
Overriding the Location of Test Source and Output.
<build> ... <testSourceDirectory>src-test</testSourceDirectory> <testOutputDirectory>classes-test</testOutputDirectory> ... </build>
Most lifecycles bind the test goal of the Surefire plugin to the test phase. The Surefire plugin is Maven’s unit testing plugin, the default behavior of Surefire is to look for all classes ending in *Test in the test source directory and to run them as JUnit tests. The Surefire plugin can also be configured to run TestNG unit tests.
After running mvn test
, you should also notice that the Surefire produces a number of reports in target/surefire-reports. This reports directory will have two files for each test executed by the Surefire plugin: an XML document containing execution information for the test, and a text file containing the output of the unit test. If there is a problem during the test phase and a unit test has failed, you can use the output of Maven and the contents of this directory to track down the cause of a test failure. This surefire-reports/ directory is also used during site generation to create an easy to read summary of all the unit tests in a project.
If you are working on a project that has some failing unit tests, but you want the project to produce output, you’ll need to configure the Surefire plugin to continue a build even if it encounters a failure. The default behavior is to stop a build whenever a unit test failure is encountered. To override this behavior, you’ll need to set the testFailureIgnore
configuration property on the Surefire plugin to true.
Configuring Surefire to Ignore Test Failures.
<build> <plugins> <plugin> <groupId>org.apache.maven.plugins</groupId> <artifactId>maven-surefire-plugin</artifactId> <configuration> <testFailureIgnore>true</testFailureIgnore> </configuration> </plugin> ... </plugins> </build>
If you would like to skip tests altogether, you can do so by executing the following command:
$ mvn install -Dmaven.test.skip=true
The maven.test.skip
variable controls both the Compiler and the Surefire plugin, if you pass in maven.test.skip
you’ve told Maven to ignore tests altogether.
The install
goal of the Install plugin is almost always bound to the install
lifecycle phase. This install:install
goal simply installs a project’s main artifact to the local repository. If you have a project with a groupId
of org.sonatype.mavenbook
, an artifactId
of simple-test
, and a version
of 1.0.2, the install:install
goal is going to copy the JAR file from target/simple-test-1.0.2.jar to ~/.m2/repository/org/sonatype/mavenbook/simple-test/1.0.2/simple-test-1.0.2.jar. If the project has POM packaging, this goal will copy the POM to the local repository.
The deploy
goal of the Deploy plugin is usually bound to the deploy
lifecycle phase. This phase is used to deploy an artifact to a remote Maven repository, this is usually required to update a remote repository when you are performing a release. The deployment procedure can be as simple as copying a file to another directory or as complex as transferring a file over SCP using a public key. Deployment settings usually involve credentials to a remote repository, and, as such, deployment settings are usually not stored in a pom.xml. Instead, deployment settings are more frequently found in an individual user’s ~/.m2/settings.xml. For now, all you need to know is that the deploy:deploy
goal is bound to the deploy
phase and it takes care of transporting an artifact to a published repository and updating any repository information which might be affected by such a deployment.