I am trying to use fsc (fast scala compiler) with my maven project. My pom.xml has:
...
<execution>
<id>cc</id>
<goals>
<goal>cc</goal>
</goals>
<phase>compile</phase>
<configuration>
<useFsc>true</useFsc>
<once>true</once>
</configuration>
</execution>
...
as discussed in What is the fastest way to compile Scala files using maven?
When I type mvn scala:cc, it hangs on:
[INFO] wait for files to compile...
Running mvn scala:cc -DdisplayCmd=true -Dverbose=true
[INFO] cmd: /bin/sh -c .../java -classpath [redacted] scala.tools.nsc.MainGenericRunner scala.tools.nsc.CompileServer >MainGenericRunner.out 2>MainGenericRunner.err
Which seems odd (shouldn't it not include scala.tools.nsc.MainGenericRunner?) I noticed that MainGenericRunner.out contains
no such file: scala.tools.nsc.CompileServer
which seems to confirm my suspicion.
Has anyone encountered this, or have a work around? I'd really like to use fsc to speed up my builds. I found one user with a similar output on google groups, but no follow ups.
Running scala 2.8.1 and maven 3.0.3 on OSX
When you invoke mvn scala:cc maven will use execution id default-cli (or something similar) which is not configured in your pom. Because of that maven will be using defaults of cc goal. Currently your pom is configured to execute cc goal with your customizations during "compile" phase. So, running something like mvn compile or mvn install should be working as you expect.
Related
When trying to build our project from within Eclipse I keep getting the following error:
Execution generate-sources of goal
org.apache.cxf:cxf-codegen-plugin:3.2.0:wsdl2java failed: A required
class was missing while executing
org.apache.cxf:cxf-codegen-plugin:3.2.0:wsdl2java:
javax/xml/bind/annotation/adapters/HexBinaryAdapter
The reason for that is that - while we still compile for a Java-8 target environment - the tool chain (i.e. Eclipse, M2E (Eclipe's Maven-plugin), Maven, and CXF) is executed using Java-11.
In Java 9+ javax/xml/bind is not part of the rt.jar anymore, hence the class is missing when the plugin tries to start up. Elsewhere I found that one can enable it by specifying an "--add-modules java.xml.bind" JVM option.
I tried adding that option to the MAVEN_OPTS environment variable but that is apparently ignored when M2E starts up Maven (and with it the CXF plugin) in a separate VM.
Next I tried to specify that option in the plugin's configuration in the pom.xml like so:
<build>
<plugins>
<plugin>
<groupId>org.apache.cxf</groupId>
<artifactId>cxf-codegen-plugin</artifactId>
<version>${cxf.version}</version>
<executions>
<execution>
<id>generate-sources</id>
<phase>generate-sources</phase>
<configuration>
<fork>true</fork>
<additionalJvmArgs>--add-modules java.xml.bind</additionalJvmArgs>
...
</configuration>
<goals>
<goal>wsdl2java</goal>
</goals>
</execution>
</executions>
...
... but that also didn't fly. :-(
Any idea anyone, how and where one can specify that option or how I can make the former standard javax-classes available to a Maven-plugin running under Java 9+ (when executed from Eclipse M2E) ?
Just in case: this is NOT an Eclipse or M2E issues! Even when I start Maven on the command line using Java 9+ I get:
...
[ERROR] Failed to execute goal org.apache.cxf:cxf-codegen-plugin:3.2.0:wsdl2java (generate-sources) on project my_project: Execution generate-sources of goal org.apache.cxf:cxf-codegen-plugin:3.2.0:wsdl2java failed: A required class was missing while executing org.apache.cxf:cxf-codegen-plugin:3.2.0:wsdl2java: javax/xml/bind/annotation/adapters/HexBinaryAdapter
When I build my project using sbt assembly, I get java.util.zip.ZipException: duplicate entry: META-INF/MANIFEST.MF. The solution that google recommends, is to use MergeStrategy.discard.
That works in getting the project to build - but it crashes at runtime, because Dropwizard (a dependency) relies on info contained in manifest.mf (full issue details: https://github.com/dropwizard/dropwizard/issues/455 ).
The recommendation when encountering that error, is to merge the manifests.
I've tried all the MergeStrategies on Manifest.MF that seem like they'd do the trick (filterDistinctLines, concat, first, last), they all cause the build to fail with java.util.zip.ZipException: duplicate entry: META-INF/MANIFEST.MF. The only thing that compiles is discard, but that causes the program to crash at runtime due to Dropwizard relying on the mf file.
Any ideas what to do here? Is there a way to merge the manifests as described in the comments at https://github.com/dropwizard/dropwizard/issues/455 ?
So I ended up doing this:
Separated the Scala & Java projects
Build the sbt project as is - copy it to the java project/lib directory
Added a system dep for lib/scala project, and used the add-jar plugin to add this jar to the classpath when building the fatjar:
<plugin>
<groupId>com.googlecode.addjars-maven-plugin</groupId>
<artifactId>addjars-maven-plugin</artifactId>
<version>1.0.5</version>
<executions>
<execution>
<goals>
<goal>add-jars</goal>
</goals>
<configuration>
<resources>
<resource>
<directory>${real.base.dir}/lib</directory>
</resource>
</resources>
</configuration>
</execution>
</executions>
</plugin>
Build the java project fatjar and deploy it - it now has access to the scala jar on the classpath.
Script (server is the name of the java project i made):
sbt clean assembly
cp target/scala*/project*.jar server/lib
cd server
mvn clean install
cd ../
cp server/target/server*.jar target
I have a set of test cases that use the Eclipse WorkbenchPage and a couple of other classes to perform a set of functions.
When I execute the test bundle in Eclipse, all the test results are green. But when I "clean install" the same package in the command prompt, the build fails and shows test failures in my test classes.
What could be the problem here? I tried debugging my code from Maven but it didn't help at all.
Tycho and Eclipse differ in how they determine the test runtime:
In Eclipse, by default the entire target platform and all projects from the workspace are included in the test runtime.
In Tycho, only the test bundle/fragment and its transitive dependencies are part of the test runtime. If your test has implicit dependencies, e.g. on a bundle which provides some UI via an extension point, you need to explicitly configure these in Tycho.
With the following build configuration, you can for example include the feature org.eclipse.rcp and all its transitive dependencies into the test runtime:
<plugin>
<groupId>org.eclipse.tycho</groupId>
<artifactId>target-platform-configuration</artifactId>
<version>${tycho-version}</version>
<configuration>
<dependency-resolution>
<extraRequirements>
<requirement>
<type>eclipse-feature</type>
<id>org.eclipse.rcp</id>
<versionRange>0.0.0</versionRange>
</requirement>
</extraRequirements>
</dependency-resolution>
</configuration>
</plugin>
I am new to maven. So I have a project with pom.xml file. So I ran that with maven and the build was successful. I have glassfish. Glassfish is already running separately. So now what is the next step to run the project with Glassfish? My IDE is eclipse.
You have to first tell Maven to build the WAR, check out this plugin for that: http://maven.apache.org/plugins/maven-war-plugin/.
Then you need to tell maven how to deploy to glassfish, you can either configure a Maven execution plugin to do this (see here: https://www.mojohaus.org/exec-maven-plugin/). Or you can look around for a custom plugin devoted to integrating maven with glassfish. This one looks promising, but I have not used it: http://maven-glassfish-plugin.java.net/.
Maven provides a lot of basic functionality out of the box, but most of the cooler stuff with build automation is done through plugins.
Update
Just updating to add a very simple Pom that will do a auto-deployment. Note: if you just run a "mvn clean install", with the packaging set to 'war', maven will build the .war file for you and place it in the target/ folder. You can take this and deploy it to glassfish manually if you just want to get started.
Below is part of a very simple pom that uses the Maven execution plugin to auto-deploy to glassfish as a function of the build:
<build>
<plugins>
<plugin>
<groupId>org.codehaus.mojo</groupId>
<artifactId>exec-maven-plugin</artifactId>
<executions>
<execution>
<goals>
<goal>exec</goal>
</goals>
<phase>install</phase>
</execution>
</executions>
<configuration>
<executable>${path-to-asadmin-util}</executable>
<arguments>
<argument>deploy</argument>
<argument>--user=${username}]</argument>
<argument>--passwordfile=${password-file}</argument>
<argument>--host=localhost</argument>
<argument>--port=4848</argument>
<argument>target/${project.name}</argument>
</arguments>
</configuration>
</plugin>
</plugins>
</build>
This basically just calls the deploy command on the glassfish asadmin utility[1]. You need to fill in the following variables:
${path-to-asadmin-util} --> this is the path to your asadmin utility
(normally in the glassfish_home/bin)
${username} --> glassfish admin username
${password-file} --> password file for logging into glassfish
admin[2]
${project.name} --> name of your war
If you want to get more complicated I suggest taking a look at this thread: GlassFish v3 and glassfish-maven-plugin (Mac).
[1] - http://docs.oracle.com/cd/E18930_01/html/821-2433/deploy-1.html#SJSASEEREFMANdeploy-1
[2] - http://docs.oracle.com/cd/E18930_01/html/821-2435/ghgrp.html#ghytn
Additonnaly, you should have a glance at this StackOverflow thread, dealing with maven deployement in glassifsh : https://stackoverflow.com/a/1836691/1047365.
For further understanding of Maven, you should REALLY read this (free) book : http://www.sonatype.com/books/mvnref-book/reference/. This is THE reference for Maven.
We can explain you what Maven is doing, producing, etc ... but Sonatype made a great work and you'll probably learn more reading it than we could ever do !
Regards.
I found this tutorial useful: http://tshikatshikaaa.blogspot.com/2012/05/introduction-to-maven-concepts-crash.html
I would like to have the maven eclipse plugin regenerate my .classpath whenever a build is run, and I did so by using the following configuration:
<!--
Generate a new .classpath each time the build is run, but don't try
to download sources or javadocs
-->
<profile>
<id>elipse-update</id>
<activation>
<file>
<exists>.classpath</exists>
</file>
</activation>
<build>
<plugins>
<plugin>
<artifactId>maven-eclipse-plugin</artifactId>
<executions>
<execution>
<phase>generate-sources</phase>
<goals>
<goal>eclipse</goal>
</goals>
<configuration>
<downloadSources>false</downloadSources>
<downloadJavadocs>false</downloadJavadocs>
</configuration>
</execution>
</executions>
</plugin>
</plugins>
</build>
</profile>
For some reason, this causes the maven jetty plugin to fail with ClassNotFoundException errors (it complains about all sorts of Spring classes not being there). Of course, it worked without a hit when I didn't have the maven eclipse plugin installed. Here's an example of what I'm talking about:
$ mvn jetty:run
[INFO] Scanning for projects...
...
[INFO] Starting jetty 6.1.22 ...
2010-02-11 20:53:08.984:INFO::jetty-6.1.22
2010-02-11 20:53:09.109:WARN::Could not instantiate listener org.springframework.web.context.ContextLoaderListener
java.lang.ClassNotFoundException: org.springframework.web.context.ContextLoaderListener
at java.net.URLClassLoader$1.run(URLClassLoader.java:200)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(URLClassLoader.java:188)
at java.lang.ClassLoader.loadClass(ClassLoader.java:306)
at org.codehaus.classworlds.RealmClassLoader.loadClassDirect(RealmClassLoader.java:195)
at org.codehaus.classworlds.DefaultClassRealm.loadClass(DefaultClassRealm.java:255)
at org.codehaus.classworlds.DefaultClassRealm.loadClass(DefaultClassRealm.java:274)
at org.codehaus.classworlds.RealmClassLoader.loadClass(RealmClassLoader.java:214)
at java.lang.ClassLoader.loadClass(ClassLoader.java:251)
Of course, if I delete that eclipse plugin section, I can run jetty as expected:
$ mvn jetty:run
[INFO] Scanning for projects...
...
Feb 11, 2010 8:55:28 PM org.springframework.web.context.ContextLoader initWebApplicationContext
INFO: Root WebApplicationContext: initialization completed in 1672 ms
2010-02-11 20:55:28.687:INFO::Started SelectChannelConnector#0.0.0.0:8080
[INFO] Started Jetty Server
[INFO] Starting scanner at interval of 5 seconds.
Some things I know:
Yes, I only activate this profile if the .classpath is present. This seems counter-intuitive, but I have a reason: I have another profile that activates when the .classpath is absent which runs the eclipse plugin with the options for downloading source code and javadocs set to true. I don't want this to happen each build, so I created a separate plugin config for when the classpath is already there.
Yes, I could simply create properties that held the values of the options I wish to change instead of specifying the entire plugin config again. In that case, I would just set a eclipse.downloadSources property to true or false depending on the presence of the classpath and have a single plugin definition in the regular build section.
Any advice? This is a strange problem.
Thanks,
LES
I suspect the Maven Eclipse Plugin to mess do some classpath woodo that affects the jetty plugin later.
But to be honest, this is not a very common way to use the Eclipse plugin, at least not to my knowledge. Most people don't include it in their build, they just run it when the POM has changed and this generally doesn't happen every build. Moreover, touching the .classpath confuses Eclipse if I remember well and forces a clean rebuild of the project which is pretty annoying.
So at the end, and I'm sorry for that, this seems to introduce more annoyances than benefits. You can try to open a Jira issue though.
You really don't want to run the maven-eclipse-plugin on every build when there's a .classpath present. I can't tell you exactly what it does to the classpaths, but this is not how it's intended to be used. The assumption is that when you run it you only run eclipse:eclipse (or some other goal).
Why do you want to keep re-running it?