Could we adjust the build to be more flexible with respect to Hadoop versions? I'm not sure how users are expected to get Hadoop setup for the book:
None of these match the book code's demand of Hadoop v1.1.1.
Allow different Hadoop versions to be specified.
See #14 and
I just pushed a change to fix this. E.g.
mvn package -Dhadoop.version=1.2.1
mvn package -Dhadoop.version=1.0.3
Sorry it took me so long - I didn't see the github issues until today.
@tomwhite No worries! Thanks for addressing this. :)
I made another modification so you can specify both the distro and the precise version (or neither). See 9ffc052
I also ran the unit and integration tests against a bunch of different versions as follows:
HADOOP_HOME=~/dev/hadoop-1.2.1 mvn clean verify -Dhadoop.version=1.2.1
HADOOP_HOME=~/dev/hadoop-0.23.10 mvn clean verify -Dhadoop.distro=apache-23 -Dhadoop.version=0.23.10
HADOOP_HOME=~/dev/hadoop-2.2.0 mvn clean verify -Dhadoop.distro=apache-2 -Dhadoop.version=2.2.0
HADOOP_HOME=~/dev/hadoop-0.20.2-cdh3u5 mvn clean verify -Dhadoop.distro=cdh3 -Dhadoop.version=0.20.2-cdh3u5
HADOOP_HOME=~/dev/hadoop-2.0.0-cdh4.5.0 mvn clean verify -Dhadoop.distro=cdh4 -Dhadoop.version=2.0.0-cdh4.5.0