Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 10 May 1999 05:45:17 +0200
From:      Eivind Eklund <eivind@FreeBSD.org>
To:        "Kenneth D. Merry" <ken@plutotech.com>
Cc:        cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/sbin/camcontrol camcontrol.8 camcontrol.c src/sys/cam cam_ccb.h cam_sim.c cam_sim.h cam_xpt.c src/sys/cam/sc
Message-ID:  <19990510054516.V76212@bitbox.follo.net>
In-Reply-To: <199905081909.NAA64618@panzer.plutotech.com>; from Kenneth D. Merry on Sat, May 08, 1999 at 01:09:36PM -0600
References:  <19990508122349.L76212@bitbox.follo.net> <199905081909.NAA64618@panzer.plutotech.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, May 08, 1999 at 01:09:36PM -0600, Kenneth D. Merry wrote:
> Eivind Eklund wrote...
> > I wasn't saying that it wasn't the right decision - I was just
> > registering surprise.
> > 
> > I think it would be appropriate to send a message to -announce, and it
> > definately needs to go into the release notes for 3.2.  Breaking
> > binary compatibility in -stable is supposed to be an event.
> 
> I don't think it's a big enough event to bother people on -announce.  If
> people are running -stable, they should read the -stable list.  I sent mail
> there, so those folks should be informed of the situation.

This does not match with the guidelines we've given to our users
previously.  We've said that they were adviced to read the -stable
list, but it wasn't a requirement for following the branch.  Adding it
as a requirement makes -stable much less useful; IMO, we don't even
*want* all users of -stable to read the -stable list, as that would
create much too much noise on the list.

I will repeat myself: Breaking binary compatibility in -stable is
*supposed to be an event*.  This means that if you are doing it, you'd
better be doing it for a reason that is good enough to send a message
to -announce.

> And this isn't the only thing that'll break binary compatibility in
> 3.2.  I fixed some problems in the devstat code back in February (just
> after 3.1 went out, I think) that will also break ports that use the
> devstat code.

Ouch.

> I think people should probably expect some interfaces to be broken between
> releases, since a completely stagnant userland<->kernel interface for
> everything would probably indicate a dead branch.

Then all of RELENG_2_2 was dead.  AFAIR, the only breaks of binary
compatibility there was ipfw (broken twice, due to people ignoring the
requirement not to break binary compatibility); at no time was
anything done that would break the installed packages.

People, we are not supposed to be creating interfaces that *need* to
break binary compatibility.  It isn't as if the techniques for
creating extensible interfaces weren't well known 20 years ago; there
is no reason not to use them today.

> It might be worth mentioning this stuff in the release notes, though.

It is 100% necessary that this be mentioned in the release notes or
backed out; if it isn't worth mentioning in the release notes, it
definately isn't worth breaking binary compatibility for.

Eivind.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990510054516.V76212>