You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
jnioche edited this page Dec 6, 2010
·
5 revisions
Prerequisites
Java 6.0
Apache ANT
Internet connection (required for fetching the dependencies with Ivy)
Compiling
Running 'ant publish-all' from the root directory of Behemoth will compile each module and a generate a jar file in the build directory of each module. The modules can have dependencies between each other (at least to gate-core) as well as external ones. IVY publishes the jars for each module locally under repository/shared/ to help resolving the dependencies between the behemoth-modules.
Testing
Running 'ant test-all' from the root directory of Behemoth calls the JUnit tests for each module. The outputs of the tests can be found in the directory build/test in the modules.
Generating a job file
A job file is necessary in order to run Behemoth on a Hadoop cluster. Job files are generated on a module basis : a user can generate several job files and use them separately (e.g. one for Tika and one for GATE) or build a new module with some custom code and declare a dependency to both the modules tika and gate and generate a job file for that new module.
cd modules/gate
ant job
will generate a behemoth-gate.job file in the build directory. This job file can then be used with Hadoop.