Mailing-List: contact hadoop-dev-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hadoop-dev@lucene.apache.org
Received-SPF: neutral (herse.apache.org: local policy)
Message-ID: <45EF11B1.4050206@apache.org>
Date: Wed, 07 Mar 2007 11:25:37 -0800
From: Doug Cutting <cutting@apache.org>
User-Agent: Thunderbird 1.5.0.10 (X11/20070306)
MIME-Version: 1.0
To: hadoop-dev@lucene.apache.org
Subject: Re: [jira] Updated: (HADOOP-1080) Cygwin path translation should
 occur earlier in bin/hadoop
References: <6932136.1173293544556.JavaMail.root@brutus>
 <45EF0D4F.3070402@getopt.org>
In-Reply-To: <45EF0D4F.3070402@getopt.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Andrzej Bialecki wrote:
> Now I'm confused ... If I were to apply this patch, should I do this 
> both in trunk/ and branches/branch-0.12 ?

You're right to be confused!

I've thus far managed merging in Hadoop lazily.  I apply everything to 
trunk, then merge things to the branch before making a branch release. 
That minimizes the number of steps, since otherwise I'd probably want to 
apply the patch to both branches, or merge as each patch is made.

I also, to simplify things, stall patches for the next major release 
while we're still expecting to make another point release.  Again, this 
simplifies merging, since all patches can be merged to the release 
branch at once, rather than patch-by-patch.

So, right now, all the 0.12.1 patches are only in trunk, and no 0.13.0 
patches are yet in trunk.  So making 0.12.1 will require a single, 
simple merge from trunk to the 0.12 branch.

Long-term, as we start releasing less frequently, maintaining branches 
longer, having more active committers, etc., this policy may not scale. 
  But it's where we are today.  Soon Nigel may take over as release 
manager, and he may propose a policy less influenced by laziness.

Doug