hadoop-common-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Harsh J <ha...@cloudera.com>
Subject Re: How to make changes in Hadoop's classes "visible"
Date Fri, 29 Apr 2011 15:57:01 GMT
Every project in the Apache Hadoop environment (Pig, Hive, etc.) have
a "tar" target in their build framework. The tar targets build a
distributable archive for the source provided (something like
downloading a release from apache's sites, but you've made it yourself
here). If you would like only the jars (which you can replace and go
ahead in most situations), I believe there's also a "jar" target in
most of these projects.

If I were to add HBASE-3714 to my HBase distribution (assuming it
applies without much trouble), I would do:

1) cd to HBASE_HOME
2) patch -p0 < ~/Path/to/diff.patch
3) mvn package [maven since HBase uses Maven AFAIK]
4) cd to target/ and locate the tar distributable file [You can use
find, etc. tools to locate it as well]

The procedure differs slightly for each of these Java projects based
on what tool they may be using as their build framework, but the idea
remains the same: patch/modify source, compile, jar/tar/package

2011/4/29 Panayotis Antonopoulos <antonopoulospan@hotmail.com>:
> Hello,
> I am a beginner and I have downloaded Hadoop 0.20.2.
> I would like to make some changes in some of the classes of Hadoop and import them into
hadoop-0.20.2-core.jar so that they are visible to the mapreduce programs.
> What is the best way to do this?
> What should I do if I find the changes at JIRA, for example if I want to change the LoadIncrementalHFiles
class as mentioned at https://issues.apache.org/jira/browse/HBASE-3714?
> Is it the same procedure?
> Thank you in advance!!
> Panagiotis Antonopoulos.

Harsh J

View raw message