Mailing-List: contact general-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: general@hadoop.apache.org
Received-SPF: neutral (athena.apache.org: local policy)
MIME-Version: 1.0
In-Reply-To: <DA9E7EF1-9A82-414E-AA5C-DD1088869CD2@apache.org>
References: <dfe484f01002161412h8bc953axee2a73d81a234bdf@mail.gmail.com>
	 <4B7D9503.4080205@apache.org>
	 <DA9E7EF1-9A82-414E-AA5C-DD1088869CD2@apache.org>
Date: Thu, 18 Feb 2010 15:06:00 -0800
Message-ID: <c9118d4c1002181506l7ec9840dq3690158ebc70c02@mail.gmail.com>
Subject: Re: Release plans
From: Jeff Hammerbacher <hammer@cloudera.com>
To: general@hadoop.apache.org
Content-Type: multipart/alternative; boundary=0023544714c0e77794047fe8025d

--0023544714c0e77794047fe8025d
Content-Type: text/plain; charset=ISO-8859-1

>
>  * Someone outside of Yahoo needs to step up and start addressing the
> blockers to get 0.21 released.
>

Thanks for the update on the state of the world at Yahoo!, Owen--very
helpful. I think the community will step up to knock down some of the
blockers once we resolve what should be in the 0.21 release: the current
branch, or a rebase on trunk. Do you/Yahoo! have a preference on that front?

Rather than set a date, I think it is time to move to feature-based
> releases. We'd need to vote on the feature set, but looking for security,
> end-to-end avro, and symlinks seems like a reasonable list. That will avoid
> the large rush of commits the last week of the deadline, which has been
> counter productive.
>

Could someone give an example of a successful open source project that
follows a feature-based release cycle? From the research I've done, the
regular drumbeat of time-based releases seem to be more conducive to project
health. Always interested to hear otherwise. The example Owen cites of "the
large rush of commits the last week of the deadline" is certainly a good
argument in favor of feature-based releases; I'm curious to hear more.

Thanks,
Jeff

--0023544714c0e77794047fe8025d--