Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 12 Feb 2004 08:41:50 -0800
From:      "Kevin Oberman" <oberman@es.net>
To:        Sergey Matveychuk <sem@ciam.ru>
Cc:        stable@freebsd.org
Subject:   Re: next release 
Message-ID:  <20040212164150.EEAB05D07@ptavv.es.net>
In-Reply-To: Message from Sergey Matveychuk <sem@ciam.ru>  of "Thu, 12 Feb 2004 07:07:47 %2B0300." <402AFC13.5010509@ciam.ru> 

next in thread | previous in thread | raw e-mail | index | archive | help
> Date: Thu, 12 Feb 2004 07:07:47 +0300
> From: Sergey Matveychuk <sem@ciam.ru>
> Sender: owner-freebsd-stable@freebsd.org
> 
> Where do you plan to release of 4.10 or 4.9.1?
> There is not info on http://www.freebsd.org/releng/index.html
> 
> I've found some problems on my home box with 4.9 and when I've upgraded 
> to RELENG-4 they are gone. But I'm afraid to install not-release on my 
> work boxes.

I understand your reluctance to install a non-release on your work
system, but the reality is that the most stable systems are about a
month AFTER a release. This makes sense if you understand how the
release process really works.

Releases are simply points along the STABLE branch that are declared to
be releases. At various times the RE declares that it's time for a
release. The STABLE branch is semi-frozen with no new features allowed
until after a release. When the word goes out the tree is about to be
frozen, lots of the developers who have been working on local stuff will
suddenly check them back into the tree. THIS IS A BAD TIME TO RUN
STABLE! (But we really appreciate those who do, as it really is
important in shaking out the bugs.)

In the month or so between the freeze and the release, a strong attempt
is made to get everything fixed up and running well. The majority of the
problems, both new and old, are fixed to the satisfaction of the RE and
a release is made. (This is vastly simplified.)

Now that the code is labeled "release", LOTS of people start to run
it. Almost always this finds new, occasionally serious bugs that nobody
reported before the release. Most of these are fixed very quickly, but
it's too late for the release, so few people run this code, even though
it is the least likely to have problems.

Now you see why I recommend that people wait for a month after a release
to update and to do it to STABLE, not release.

I have no official position in FreeBSD organization and speak only for
myself, not FreeBSD or my employer. Core and RE may completely think I
am confused and/or wrong about this, but I see the cycle every
release. And I don't have a "fix" as the cause is really human nature.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman@es.net			Phone: +1 510 486-8634



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040212164150.EEAB05D07>