Date: Sat, 07 Jun 2008 19:47:33 -0400 From: Ken Smith <kensmith@cse.Buffalo.EDU> To: Jo Rhett <jrhett@netconsonance.com> Cc: freebsd-stable <freebsd-stable@freebsd.org> Subject: Re: challenge: end of life for 6.2 is premature with buggy 6.3 Message-ID: <1212882453.94878.51.camel@neo.cse.buffalo.edu> In-Reply-To: <52E265BB-BBB1-439D-B375-D6C6AA04697C@netconsonance.com> References: <9B7FE91B-9C2E-4732-866C-930AC6022A40@netconsonance.com> <48472DB6.5030909@samsco.org> <6010676B-91B0-4AF8-ACF8-039A59B29331@netconsonance.com> <200806050248.59229.max@love2party.net> <B9B83C12-7130-490D-A4BE-0469711B24DC@netconsonance.com> <20080605083907.GD1028@server.vk2pj.dyndns.org> <902E9703E6E50776A17E9F92@utd65257.utdallas.edu> <20080605220244.GP1028@server.vk2pj.dyndns.org> <34E9F0D46D7B9F45EDA38F4C@utd65257.utdallas.edu> <48488D1A.9070105@kutulu.org> <52E265BB-BBB1-439D-B375-D6C6AA04697C@netconsonance.com>
next in thread | previous in thread | raw e-mail | index | archive | help
--=-bnqouZLPRbcNqFe/S59c
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable
On Sat, 2008-06-07 at 14:37 -0700, Jo Rhett wrote:
> These are the raw issues without any friendly wording.
>=20
> 1. Bugs in 6.3 that are patched aren't available in any other -RELEASE.
> 2. Bugs in 6.3 outstanding that don't affect 6.2
> 3. Overall amount of bugs.
> 4. Difference in code base between 6.3 and 6-STABLE is > than 6.2 and =20
> 6.3
>=20
> These combine to produce a release which will never be "stable" for =20
> production needs.
The issue seems to be that we see no evidence that there is any
significant volume to your item #2, and that's why you are receiving so
much push-back on this. And that's what you are being asked for
evidence of. That is the ONLY thing that should be preventing you from
upgrading to 6.3. Item #1 can't be addressed at all and affects you
while you're running 6.2 just as much as it would affect you running
6.3. Item #3 is irrelevant in most peoples' eyes because again we see
no evidence that those bugs that would be part of #3 are not in 6.2 as
well as 6.3. Item #4 is also a non-issue since there is no telling if
any of that code that's changed affects something that you would notice.
For example if a chunk of the code that's changed between 6.3 and
6-STABLE were an amd(8) import that has absolutely no relevance to any
of this discussion, 6.3's overall stability, etc.
As for your overall question of "Why can't 6.2 continue to be
supported?" I answered that but from your reply it seems you may not
have quite understood it. We typically support the LAST release in a
branch for 2 years, it's typical to not support the earlier releases
that long. That is based on the theory that we don't have regressions,
and that if we do have regressions we fix them with Errata Notices. We
did indeed have a major regression with 6.3 - there was a problem with
the threads libraries which has been fixed with an Errata Notice.
As I say the reason for so much push-back seems to be the lack of any
concrete evidence that there are major regressions (by definition a
regression being something that had worked in 6.2 that is broken in
6.3). From peoples' experiences and their surveys of the PR list we
have not seen any evidence of the regressions you say you are concerned
about - it seems like anything that would affect you in 6.3 should also
be affecting you in 6.2. Since we all seem to be unable to find what
you are concerned about could you please (when you have time - I have
noticed you said you were time-crunched) let us know what they are?
Given that perhaps we can address the issues and do Errata Notices.
You have also asked about costs. A large one is the time of the
volunteers on the security team. We take Security Advisories and Errata
Notices VERY seriously - we try very hard to make sure that what gets
posted is absolutely correct. That means applying the patches to all
branches concerned and testing them as best we can. As code ages some
of the more complicated patches can cause concern that they're not quite
right for the older code because of fundamental changes that have been
made (for example locking kernel data structures is hard, at times
extremely complex, and what you need to "watch out for" can change over
time even within a X-STABLE branch). And there is some cost in
time/energy/resources to maintain FreeBSD-Update support.
Those things don't mean we can't continue support of 6.2, but they do
mean it requires effort. And again we can't seem to find any evidence
that the effort is needed since we can't seem to find any evidence of
*regressions* in 6.3. So when you've got the time let us know what
evidence it is we're missing. And again, pointing at the CVS repository
and saying "lots changed" isn't evidence. That's completely irrelevant
because of many things including what I mentioned above (e.g. an import
of vendor code shows up as changed code in our CVS repository but for
the purposes of this discussion it's irrelevant).
--=20
Ken Smith
- From there to here, from here to | kensmith@cse.buffalo.edu
there, funny things are everywhere. |
- Theodore Geisel |
--=-bnqouZLPRbcNqFe/S59c
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (FreeBSD)
iEYEABECAAYFAkhLHhUACgkQ/G14VSmup/Z03wCdHckc5laQbiJ2mo/di+Nb5EfF
sVEAn0gb6o7ZA80BHjtcnZdDUe6iDlLB
=EeH+
-----END PGP SIGNATURE-----
--=-bnqouZLPRbcNqFe/S59c--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1212882453.94878.51.camel>