[
https://issues.apache.org/jira/browse/PIG-3892?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Daniel Dai updated PIG-3892:
----------------------------
Attachment: PIG-3892-1.patch
Attach patch. Include several changes:
1. A new ant target "jar-all-h12" which build pig-withouthadoop-h1.jar,
pig-withouthadoop-h2.jar, pig-h1.jar. Intentionally skip pig-h2.jar since one
fat pig.jar should be enough
2. "jar-all-h12" is only invoked when doing release (ant package/tar)
3. "jar/jar-withouthadoop/jar-all" targets remains the same, except for adding
h1/h2 suffix according to hadoopversion
4. maven artifacts remain the same (no suffix, h2 classifier for h2 artifacts)
5. pig script will chose h1/h2 jar based on existence of hadoop-core.jar
(hadoop-core.jar indicates h1)
> Pig distribution for hadoop 2
> -----------------------------
>
> Key: PIG-3892
> URL: https://issues.apache.org/jira/browse/PIG-3892
> Project: Pig
> Issue Type: Bug
> Components: build
> Reporter: Daniel Dai
> Assignee: Daniel Dai
> Fix For: 0.13.0
>
> Attachments: PIG-3892-1.patch
>
>
> Currently Pig distribution only bundle pig.jar for Hadoop 1. For Hadoop 2
> users they need to compile again using -Dhadoopversion=23 flag. That is a
> quite confusing process. We need to make Pig work with Hadoop 2 out of box. I
> am thinking two approaches:
> 1. Bundle both pig-h1.jar and pig-h2.jar in distribution, and bin/pig will
> chose the right pig.jar to run
> 2. Make two Pig distributions for Hadoop 1 and Hadoop
> Any opinion?
--
This message was sent by Atlassian JIRA
(v6.2#6252)
