Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
Hadoop, mail # dev - After Creating a patch what I need to do


+
Thoihen Maibam 2013-05-01, 13:33
+
Steve Loughran 2013-05-01, 16:40
Copy link to this message
-
Re: After Creating a patch what I need to do
Thoihen Maibam 2013-05-01, 18:45
@Steve thanks a lot specially that Linux VM with virtualbox that gave me
some idea how to test it.
On Wed, May 1, 2013 at 10:10 PM, Steve Loughran <[EMAIL PROTECTED]>wrote:

> On 1 May 2013 06:33, Thoihen Maibam <[EMAIL PROTECTED]> wrote:
>
> > Hi All,
> >
> > Can somebody help me after creating patch what I need to I do. I have
> seen
> > one subject 'How to test the patch' but that isn't really helping me. I
> am
> > stuck in the below areas.
> >
> > dev-support/test-patch can test your patch
>
> or you can cheat and just upload the patch to jenkins and wait to get told
> off by mail.
>
> do make sure that a full test run, mvn test, works for everything first. In
> fact, make sure that works on your machine on the unpatched trunk, as that
> way if it fails, you can be confident that it is your machine setup, not
> your patch itself.
>
> 1. I saw findbugs and forest (I don't know much about this applications
> > where to put the jars)
> >
>
>
>
>
> > 2. Do I have to upload both the HADOOP-X.patch and patch.txt(I have
> > observed this file which contains the test class) . Here why the file
> > extension is txt.
> >
>
> the .patch file must include everything.
>
> If you develop on git on a private branch try
>
> git diff trunk...HEAD > HADOOP-1234.patch
>
> this diff looks at everything that changed between when you forked off
> trunk and the patch. for long-lived branches this isn't enough, as trunk
> changes my be incompatible. For that, git rebasing helps
>
>
> >
> > 3. Can I test the patch in standalone mode as I do not have access to
> > cluster.
> >
> >
> I'd recommend creating a Linux VM with virtualbox or similar, using it to
> test on Linux -if you create >1 then you can actually bring up a virtual
> cluster.
>
> I personally use VMs running remotely on Rackspace infrastructure (CoI
> statement: I'm working with them), because it lets me schedule overnight
> test runs while I switch off the laptop and do other things. Then I get to
> find out what failed in the morning.
>
> -steve
>