IF u want to compile a hbase project specified to your own hardware configures(eg. 32 bit jdk to 64),OR release after some changes eg. applied a patch ,optimized some features,u can try this one case.
there are some rough steps to compile a hbase project:
1.download a tarball
a.u can down it direclty by access apache site ,or
b.use git to download
2.use mvn or IDE to compile
a.after downloading a raw hbase tarball(if use 1.a),untar it to a dir like hbase-0.94.24.
b.follow the commands in apache book
for instance,u can use this to generate a final deployable tarball:
mvn -DskipTests clean install && mvn -DskipTests package assembly:single
after this,u can find some hbase-0.94.24.jar ,hbase-0.94.24.tar.gz under the target dir.untar the later,u will see a final deployable files.
note:with this version,i find a test error in TestRegionObserverScannerOpenHook.java when use
mvn test -P runSmallTests
it complains that null pointer excpetion when executes MiniDFSCluster#startDataNodes(xx...).but if i run it in my eclipse ,it passes without any failures.so u can mv this java test file out of src/test/.... to anywhere.
IDE compile
a.build a mvn project that eclipse will recognize by plugin m2e which must be installed previously:
mvn eclipse:eclipse
b.open eclipse,create a project by choosing 'General' -> 'Existing proect into workspace',than find the target mvn project dir
c.after above steps, u can compile hbase by using eigther eclipse or mvn
others
of course ,with mvn, u can install snappy compress alogrithm also,see apache book
last,u will see there is a dir named '.m2' lied on ~ dir which contains some jar filses,e.g hadoop-test jar ..(downloaded by mvn automatically)