My project uses YAJSW wrapper. When i build project, i simple copy jars, required by YAJSW into assembly. So YAJSW's jars simple commited into my svn. I don't like this because my project hold 25 MB of space and 20 MB of that is YAJSW's libs.
I want to try:
delete unnecessary YAJSW's libs, but i don't know which of them are required.
download required libs from repo during build. But how to make it easy without listing all YAJSW jars in dependencies?
It will be best if there is a YAJSW pom with dependencies, but i don't find that. Maybe someone create that, so i can upload it to my repo?
This is most fundamental things you need to copy in your assembly cycle:
<fileSet>
<directory>
your_yajsw_folder
</directory>
<outputDirectory>/yajsw</outputDirectory>
<includes>
<include>bat/</include>
<include>conf/</include>
<include>lib/</include>
<include>scripts/</include>
<include>wrapper.jar</include>
<include>wrapperApp.jar</include>
</includes>
</fileSet>
Right now I'm trying to add the wrapper.jar and wrapperApp.jar from maven, hence removing it from the assymbly file. However for that I might need to edit the scripts to look for the jar in the lib folder instead of the wrapper_home
I found the maven dependencies:
<dependency>
<groupId>org.rzo.yajsw</groupId>
<artifactId>wrapper</artifactId>
<version>11.11</version>
</dependency>
<dependency>
<groupId>org.rzo.yajsw</groupId>
<artifactId>wrapperApp</artifactId>
<version>11.11</version>
</dependency>
Indeed, I have done this - please check out my project at:
https://github.com/griels/yajsw-maven-mk2
I think there are some other projects that have performed such an integration, e.g.
https://github.com/neo4j-attic/yajsw
But I couldn't find much info on them.
Alternatively, if you can find some way to run the YAJSW Gradle build scripts from Maven, you might be able to use that. I had to track down most of the dependencies by hand.
Related
I have a java spark maven project. I am building an uber jar, this jar size is huge. I dont think I am using all of the dependencies it is packaging. My questions:
How do I find out the jars(dependencies) that are not used during compile/run?
How do I find out what is 'provided' vs the ones I need to package?
I see some poms having the following to do exactly that. How do I find out this list for my use case? Is this trial and error? Like I remove some, deploy and test? somehow even with a handful of entries in my pom, somehow maven has imported about 2000 jars. How do people find out the absolutely needed set of jars and keep the size to as minimum as possible?
<artifactSet>
<excludes>
<exclude>org.bouncycastle:bcpkix-jdk15on</exclude>
<exclude>org.bouncycastle:bcprov-jdk15on</exclude>
<exclude>com.google.code.findbugs:jsr305</exclude>
<exclude>com.apple:AppleJavaExtensions:*</exclude>
<exclude>javax.servlet:*</exclude>
<exclude>org.apache.hadoop:*</exclude>
<!-- Other long list of jars -->
</excludes>
</artifactSet>
Expected to build a small jar
I'm using jar-with-dependencies in Maven to create a compiled jar file for my project. However, one of the dependencies has got source code in its jar in a non-standard place (/jvm/), and this is being included in the combined output jar.
How do I exclude a specific directory in a specific jar from being included in the combined output jar? An <excludes> tag doesn't seem to be valid anywhere in the plugin definition (at least, IntelliJ doesn't think it's valid)
I have never used jar-with-dependencies but for the same problem you mentioned above we used maven-shade-plugin in our project to include and exclude classes from particular packages.
<excludes>
<exclude>com.techidiocy.dump.*</exclude>
</excludes>
In above it will ignore all the classes from the package com.techidiocy.dump
Thanks
Why is it that although I've
set the proper path to the JBoss EJB deployment descriptor in my project's EJB facet
added jboss-ejb3.xml to Intellij's artifact Patrac-web:war exploded's <output root>/WEB-INF
that any time I make the simplest change to pom.xml Intellij removes the following entry from ${PROJECT_DIR}/.idea/artifacts/Patrac_web_war_exploded.xml:
<element id="file-copy" path="$PROJECT_DIR$/Patrac-ejb/src/main/resources/META-INF/jboss-ejb3.xml" />
and, as a result, jboss-ejb3.xml does not get copied to the target directory?
It's as though each time I make a change to pom.xml Intellij "reloads" the deployment configuration using the POM to override what settings I make within the IDE. Perhaps because I have no entry in my pom.xml for copying jboss-ejb3.xml from source directory to target directory the settings I make in Intellij IDE keep disappearing whenever Intellij "reloads." Pure conjecture on my part, but this is what seems to be happening.
If so, what change do I need to make to pom.xml in order to make this stop happening?
When a project is (re)imported from Maven IDEA configures it such way that when you invoke 'Build' from IDEA it produces the same result as Maven's 'package' goal. If you need to copy jboss-ejb3.xml to WEB-INF just put it under 'src/main/webapp/WEB-INF' directory and it will be copied by Maven and so do IDEA.
Here is an alternative to Nik's solution that I tried because I wanted to leave jboss-ejb3.xml in META-INF. Taking a look at the maven documentation, which shows how to treat jboss-ejb3.xml as a web resource and copy it into WEB-INF, I added the following to the maven-war-plugin configuration and the problem was resolved. Well, kinda sorta. But not really.
<webResources>
<resource>
<directory>../Patrac-ejb/src/main/resources/META-INF</directory>
<includes>
<include>jboss-ejb3.xml</include>
</includes>
<targetPath>WEB-INF</targetPath>
</resource>
</webResources>
Although doing it this way eliminated the need to fiddle around with IDEA facet configuration settings (because Maven was configured to copy the file, not IDEA), a new problem was introduced: two copies of jboss-ejb3.xml appeared in the WAR, one in WEB-INF and the other inside the EJB JAR (in META-INF). Although there were no apparent consequences (the application ran just fine) I preferred not to have a duplicate copy of the descriptor located inside the EJB JAR.
Using the Maven EJB plugin documentation, I tried to add an exclusion to my maven-ejb-plugin configuration e.g.
<configuration>
<ejbVersion>3.1</ejbVersion>
<excludes>
<exclude>**/jboss-ejb3.xml</exclude>
</excludes>
</configuration>
This should have prevented the duplicate copy of jboss-ejb3.xml from appearing in the EJB JAR META-INF but it didn't work for me, and after fruitlessly googling various variations of "maven-ejb-plugin excludes not working properly" I gave up.
If I could have gotten the excludes to work then I would have preferred this solution over moving jboss-ejb3.xml into src/main/webapp/WEB-INF because although this solution is slightly more complex (it requires additional Maven configuration settings in two POMs), the EJB-related descriptor would remain in the EJB module.
I've decided to go with Nik's solution until I can resolve the excludes problem.
I am using Netbeans to build a Maven project, and have the JTidy java library as a dependency. It turns out JTidy doesnt exist in any maven repos, so I can't just add a "normal" depedency entry for it.
What is the best way of handling dependencies to libraries in Maven projects that arent available on repos?
I've currently tried adding it to my maven pom as such (after copying the jar to my projects /libs folder)
<dependency>
<groupId>org.w3c</groupId>
<artifactId>org.w3c.tidy</artifactId>
<version>9.3.8</version>
<scope>system</scope>
<systemPath>${basedir}/libs/jtidy-r938.jar</systemPath>
</dependency>
However it complains that it will be unresolvable by dependent projects.
First of all, it's under another groupId, that's why you didn't find it.
<dependency>
<groupId>net.sf.jtidy</groupId>
<artifactId>jtidy</artifactId>
<version>r938</version>
</dependency>
Jtidy
But to answer your question, one way of doing this is to manually install it in your local repo as described here.
The best way IMHO is to add it to a proxy like Nexus. That way other people can access it from there without having to install it locally. However, this means you have to set up a repository manager, which doesn't make much sense if you are the only developer on the project.
We are adding new code to an existing project that uses a custom build system developed with Ant and Ivy for dependency management.
Our new team is used to Maven and its features like testing execution, cobertura reports, etc.
Our question is: is it viable to add a pom.xml matching the current project structure, but instruct Maven to load its classpath from the "lib" dir already filled by Ivy?
In other words: we want to use Maven without its dependency management.
One really dirty approach would be to generate one big jar from the libdir and config the pom.xml to include just that... but we believe there should be cleaner approach.
Any idea or recommendation?
Note: we are not interested in generating a pom.xml with dependencies from the Ivy config, we just want Maven to rely on Ivy's generated classpath. No need to discriminate between test/runtime/compile classpath.
This is our final setup to solve this:
For each Ivy legacy project, use ivy:makepom and manual inspection to figure out the dependencies that we need to send to the new projects (Maven-based). This is a one-time process for each project.
Modify the legacy build system in a way that, every time a project is built, the identified dependencies are also exported to a mvn repo. Because de build machine holds the internal repo, we just use mvn install.
In the new maven projects, declare each dependency in the pom.xml and make sure the build system runs maven builds after the legacy builds.
Thank you all for your help!
One possibility is to use the system scope to define your dependencies in maven. This allows maven to use the jars downloaded by ivy for its dependencies.
e.g.
<dependencies>
<dependency>
<groupId>group.id</groupId>
<artifactId>artifact</artifactId>
<version>a.b.c</version>
<scope>system</scope>
<systemPath>${basedir}/lib/artifact-a.b.c.jar</systemPath>
</dependency>
...
</dependencies>
Maybe the makepom task will be helpful, it creates a pom from the ivy file.
Example from that page:
<ivy:makepom ivyfile="${basedir}/path/to/ivy.xml" pomfile="${basedir}/path/to/module.pom" conf="default,runtime">
<mapping conf="default" scope="compile"/>
<mapping conf="runtime" scope="runtime"/>
<dependency group="com.acme" artifact="acme-logging" version="1.0" optional="true"/>
</ivy:makepom>