buildr looks like a very capable tool for building Java programs, but I have had a lot of trouble with using it for very simple Hadoop jobs. I discussed this on IRC a little yesterday, but wasn't able to make much progress. Full disclosure: I haven't done a lot of Java development, and haven't use Ant, Maven, or any other Java build tool (I am trying to avoid using them, they look pretty bad).
I am using Cloudera's CDH4 Hadoop distribution, installed via packages for Ubuntu 12.04. Cloudera provides a Maven repository [1] which I am using; the artifact downloads successfully. The JAR seems to contain only a pom.properties and pom.xml. The actual .jar files containing the classes are in /usr/lib/hadoop/client-0.20, which is specified via the $CLASSPATH environment variable for manual compilation with javac. When I run `buildr compile`, I get errors similar to those I got when compiling with javac without the $CLASSPATH properly set. For example: src/main/java/AvgDist.java:6: package org.apache.hadoop.fs does not exist import org.apache.hadoop.fs.Path; My buildfile is attached. Chris Guirl [1] https://ccp.cloudera.com/display/CDH4B2/Using+the+CDH4+Maven+Repository
