Prerequisites:
- A Java 8-compatible (or higher) Java development kit (JDK).
- If
JAVA_HOMEis not set on Mac OS X:export JAVA_HOME=$(/usr/libexec/java_home)
- If
- Maven 3.3.0+ (Maven Wrapper provided with
./mvnw) - Apache Ant 1.8+ (see jruby#2236)
- Make and a C++ compiler for installing the jruby-launcher gem
JRuby uses Maven for building and bootstrapping itself, along with Rake, RSpec, and MSpec for running integration tests.
The first time you enter a new source dump of JRuby (from a src zip or from a git clone), you need to build the lib/jruby.jar. The command to execute is:
./mvnw
This will run the default "install" goal (mvn install) and will do all of the following:
- Compile JRuby
- Build
lib/jruby.jar, needed for running at command line - It will install the default gems specifications
lib/ruby/gems/shared/specifications/default/and the ruby files of those gems inlib/ruby/stdlib/.
The environment is now suitable for running Ruby applications.
Note: RVM users must first run:
rvm use system
to make sure you do not use another Ruby's gems or execute another Ruby implementation.
After building lib/jruby.jar, JRuby can be run with the bin/jruby executable. If the jruby-launcher gem installed successfully, this will be a native
executable for your platform; otherwise, it will be a copy of the
bin/jruby.bash bash script.
RubyGems is installed by default, and available in bin/gem. It will
attempt to locate the jruby executable using /usr/bin/env, so you
will need the bin dir in your PATH environment or you will need to
call it via JRuby using jruby -S gem ....
The -S flag will run any script installed in JRuby's bin dir by RubyGems.
This can be a simple way to ensure you're running the JRuby (or Ruby) version
you think you are.
JRuby employs a large suite of tests, so there are many ways you can verify that JRuby is still fully functional.
for a general overview of the different directories and maven artifacts see JRuby Build
For this only the pom.rb needs to edited. using mvn-3.3.x or the maven wrapper ./mvnw will generate the pom.xml file where needed. For the jar files of the build those pom.xml will be generated for some use-cases, i.e. some IDEs need them.
To regenerate the pom.xml just run ./mvnw which will create them.
In order to prepare JRuby for testing, you must bootstrap the dev environment. This will do the following:
- Install rspec, rake, minitest, minitest-excludes, and dependencies needed to run integration tests.
mvn -Pbootstrap
In case there is a problem with installing the jruby-launcher (due to missing compiler or so) use
mvn -Pbootstrap-no-launcher
This only needs to be run once to install these gems or if you update one of the gems to a newer version or clean out all installed gems.
After changing Java code, you can recompile quickly by running one of the jar files by
mvn -pl core
For normal day-to-day testing, we recommend running the Ruby specs. We have set aside a "fast" grouping that takes only a couple minutes to run:
jruby -S rake spec:ruby:fast
For a more intensive workout, you can also run the Ruby (MRI) tests via the following rake command:
jruby -S rake test:mri
This suite takes a while to complete, so if you want to run an individual file from MRI's tests (under test/mri), use one of the following commands:
The MRI suite (under test/mri) has a runner script in test/mri/runner.rb that sets up
an appropriate test environment. Many of the MRI tests will need to be run via this script.
jruby test/mri/runner.rb test/mri/<path to test>
You can pass -v to the runner for verbose output or -n test_method_name to only run a single test method.
The runner script provides a mechanism for "excluding" known failing tests. These are usually features that JRuby has not yet implemented or can't implement on the JVM.
Excludes are in the form of Ruby scripts under test/mri/exclude, named based on the name of the test case's class, exclude with comment tests known to fail.
To run a given test with these excludes enabled, you can use the --excludes flag:
bin/jruby test/mri/runner.rb --excludes=test/mri/excludes <test file>
Most of the specs under the spec/ directory are written for rspec, and can be run with rspec.
The notable exception is the "Ruby specs" under spec/ruby, which are run with mspec as described later in this document.
rspec will be installed with mvn package -Pbootstrap or you can install it manually.
jruby -S rspec spec/path/to/spec
The specs under spec/ruby are part of the "Ruby spec" suite of tests and use the "mspec" tool to run.
Individual specs can be run with the mspec tool:
jruby spec/mspec/bin/mspec ci spec/ruby/<path to spec>
If ci is omitted or replaced with run you will see any specs known to fail. The ci command
avoids running those specs.
If you are familiar with Java debuggers, you can attach one to a JRuby process using the JDWP agent. The exact flag may vary with debugger and platform:
JRUBY_OPTS="-J-agentlib:jdwp=transport=dt_socket,server=y,suspend=y,address=localhost:5005" bin/jruby <rest of arguments>
If you are making changes that would affect JRuby's core runtime or embedding APIs, you should run JRuby's Java-based unit tests via
mvn -Ptest
There are some maven integration tests (i.e. consistency test if all gems are included, osgi test, etc) for the various distributions of JRuby which can be invoked with
mvn -Pmain -Dinvoker.skip=false
mvn -Pcomplete -Dinvoker.skip=false
mvn -Pdist -Dinvoker.skip=false
JRuby runs CI tests on TravisCI. See .travis.yml.
maven integration tests will use the packed maven artifact to run the tests in a forked maven instance. These maven projects are locatated in
maven/jruby/src/it
maven/jruby-complete/src/it
maven/jruby-jars/src/it
maven/jruby-dist/src/it
To trigger the tests with the build:
mvn -Pmain -Dinvoker.skip=false
mvn -Pcomplete -Dinvoker.skip=false
mvn -Pdist -Dinvoker.skip=false
mvn -Pjruby-jars -Dinvoker.skip=false
To pick a particular test, add the name of the directory inside the respective src/it folder, like (wildcards are possible):
mvn -Pmain -Dinvoker.skip=false -Dinvoker.test=integrity
mvn -Pmain -Dinvoker.skip=false -Dinvoker.test=j2ee*
mvn -Pmain -Dinvoker.skip=false -Dinvoker.test=osgi*
To clean the build it is important to use the same profile for the clean as what you want to build. The best way to clean build something is, i.e. jruby-jars
mvn clean install -Pjruby-jars
This first cleans everything and then starts the new build in one go!
Cleaning the build may be necessary after switching to a different version of JRuby (for example, after switching git branches) to ensure that everything is rebuilt properly.
NOTE: mvn clean just cleans the jruby-core artifact and the ./lib/jruby.jar!
Clean everything:
mvn -Pclean
All distribution packages need maven-3.3.x or the use of supplied maven wrapper. All examples below will show the use of the maven wrapper.
./mvnw -Pdist
The files will be in ./maven/jruby-dist/target.
./mvnw -Pcomplete
The file will be in ./maven/jruby-complete/target.
./mvnw -Pmain
And those files will be installed in your maven local-repository ready to use with maven, ivy, buildr, etc.
./mvnw -Pjruby-jars
The gem will be in ./maven/jruby-jars/pkg.
./mvnw -Pall
This will also clean the ext directories, i.e. a new build will then use the latest code from there for lib/ruby.
./mvnw -Pclean
First set the new version in the file VERSION inside the root directory and then to deploy the maven artifact to sonatype oss execute:
./mvnw clean deploy -Psonatype-oss-release
Go to https://oss.sonatype.org/ and close the deployment, which will check if all 'required' files are in place and then finally push the release to Maven Central and . . .
After the release, set the new development version in VERSION and generate the pom.xml files:
./mvnw