Mailing-List: contact dev-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Received-SPF: neutral (nike.apache.org: local policy)
Subject: Re: Issue involving non-sun java build
From: Renata Ghisloti Duarte de Souza <rgduarte@linux.vnet.ibm.com>
To: dev@hive.apache.org
In-Reply-To: <20130227191400.GN22265@Odie>
References: <1361981171.13492.11.camel@oc5123520263.ibm.com>
	 <CAFukC=4tPSWQ7pnN=i6PQk_xPe2P9S5O0xE8brTtSNbzD3gkQw@mail.gmail.com>
	 <20130227191400.GN22265@Odie>
Content-Type: text/plain; charset="UTF-8"
Date: Mon, 25 Mar 2013 14:00:06 -0300
Message-ID: <1364230807.11899.0.camel@oc5123520263.ibm.com>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit

Hello Brock and Jarcec,

Thank you very much for your answer!

Brock, thanks for the link. It was very helpful.

Jarcec, yes I agree that Hadoop itself still has some problems with
non-Sun Java as well. Nevertheless, I believe there are some people
working to improve that, and hopefully, 
soon Hive and Hadoop will run with all JVMs. :)

I have started contributing with some patches:

https://issues.apache.org/jira/browse/HIVE-3989
https://issues.apache.org/jira/browse/HIVE-3691
https://issues.apache.org/jira/browse/HIVE-3812


Thanks,

Renata.

On Wed, 2013-02-27 at 11:14 -0800, Jarek Jarcec Cecho wrote:
> Hi Renata,
> I'm glad to see you interested in contributing to Hive community! Please don't hesitate and follow the link provided by Brock in previous email if you're interested.
> 
> I just wanted to add that Hive is a Hadoop SQL Engine and thus we have major dependencies on Hadoop. Whereas extending Hive ability to work on other JDKs is definitely a great thing to do, I feel the need to warn that you might get into issues as Hadoop itself might not work on those JDKs. I know about following wiki page [1] that describes the support of various JDKs in Hadoop, but it seems not maintained any more.
> 
> Jarcec
> 
> Links:
> 1: http://wiki.apache.org/hadoop/HadoopJavaVersions
> 
> On Wed, Feb 27, 2013 at 11:01:58AM -0600, Brock Noland wrote:
> > Hi,
> > 
> > I think we'd want hive to work on as many JVM's as feasible. With that
> > said, since it's tested mostly on Sun JVM, it's possible we'll introduce
> > new issues in the future so you'll need to keep testing. Here is a guide on
> > how to contribute:
> > 
> > https://cwiki.apache.org/confluence/display/Hive/HowToContribute
> > 
> > Glad to have you interested!
> > Brock
> > 
> > 
> > On Wed, Feb 27, 2013 at 10:06 AM, Renata Ghisloti Duarte de Souza <
> > rgduarte@linux.vnet.ibm.com> wrote:
> > 
> > > Hello,
> > >
> > > I've been working with Hive for a couple of months now, and although it
> > > is a pretty strong framework, I can't help but noticing that some
> > > testcases fails on Non-Sun Java.
> > >
> > > Some examples are: TestCliDriver, TestParse and TestJdbcDriver.
> > >
> > > The issues involve HashMap situation, where Sun Java has a different
> > > output order than Non-Sun Java.  It is a silly problem, but it does
> > > cause failures.
> > >
> > > I've been working on fixes for theses problems, and was planning to
> > > contribute it. Is this something the Hive community would be interested
> > > in? What are your thoughts about that?
> > >
> > > Thanks,
> > >
> > >
> > > Renata.
> > >
> > >
> > 
> > 
> > -- 
> > Apache MRUnit - Unit testing MapReduce - http://incubator.apache.org/mrunit/