Re: rpm-3.0.7?

[Jeff Johnson <jbj JBJ ORG>]

On Fri, Jun 29, 2001 at 03:52:13PM -0500, Jim Knoble wrote:
> Here's a sequence of message excerpts from rpm-list:
> 
> ----------------
> Date: Tue, 19 Sep 2000 17:56:11 -0400
> From: Jeff Johnson <jbj@JBJ.ORG>
> To: rpm-list@redhat.com
> Subject: Re: glibc 2.1.3 vs. RPM 3.0.6 (was Re: glibc - update)
> Message-ID: <20000919175611.F8135@eastham.redhat.com>
> 
>   [...]
>: There's a typo in the spec file in rpm-3.0.6. That means that
>: 	Conflicts: glibc >= 2.1.3
>: should have been
>: 	Conflicts: glibc > 2.1.3
>: 
>: FWIW, the dependency is there to prevent rpm from being installed on a system
>: with glibc-2.2 that has an incompatible libio interface.
>: 
>: Install with --nodeps until rpm-3.0.7 is released to fix the typo.
>: 
>: 73 de Jeff
> 
> ----------------
> Date: Thu, 19 Oct 2000 23:08:33 -0400
> From: Jeff Johnson <jbj@JBJ.ORG>
> To: rpm-list@redhat.com
> Subject: Re: rpm versions
> Message-ID: <20001019230833.W12574@eastham.redhat.com>
> 
>   [...]
>: The recommended solution is to update rpm to the latest released
>: "production" version available through errata/updates. This is
>: currently rpm-3.0.5, soon to become rpm-3.0.7 when a new version
>: of up2date is released (python binding changes will be needed),
>: and will probably be rpm-4.0 in a couple months or so.
>: 
>: 73 de Jeff
> 
> ----------------
> Date: Fri, 22 Dec 2000 14:24:52 -0500
> From: Jeff Johnson <jbj@JBJ.ORG>
> To: rpm-list@redhat.com
> Subject: rpm-4.0.1 available for testing
> Message-ID: <20001222142452.H469@eastham.redhat.com>
> 
>: I've just put up a preliminary version of rpm-4.0.1 for testing at
>   [...]
>: 
>: A short note on rpm versions:
>: 
>: rpm 3.0.7 is the current "production" version of rpm which is going to be
>: released as an errata Real Soon Now. This is end of life form rpm-3.0.x,
>: although I will continue to add critical bug fixes in CVS until the dust
>: clears.
>   [...]
>: 
>: 73 de Jeff
> 
> ----------------
> Date: Mon, 19 Feb 2001 17:04:12 -0500
> From: Jeff Johnson <jbj@JBJ.ORG>
> To: rpm-list@redhat.com
> Subject: rpm-4.0.2 release candidate
> Message-ID: <20010219170412.X482@eastham.redhat.com>
> 
>: I've just put up a version of rpm-4.0.2 for testing at
>   [...]
>: 
>: A short note on rpm versions:
>: 
>: rpm 3.0.7 is the current "production" version of rpm which is going to be
>: replaced by rpm-4.0.2 as an errata Real Soon Now. This is end of life for
>: rpm-3.0.x, R.I.P.
>   [...]
>: 
>: 73 de Jeff
> 
> ----------------
> From: Jeff Johnson <jbj@JBJ.ORG>
> To: rpm-list@redhat.com
> Subject: Re: [Red-carpet] fix for red-carpet 0.9.1 w/ rpm 4.0.2 on RH 6.2
> Message-ID: <20010306124737.M482@eastham.redhat.com>
> Date: Tue, 6 Mar 2001 12:47:37 -0500
> 
>   [...]
>: Yup. Two ways out, both of which have the essential fix, a headerSort during
>: headerLoad.
>: 	a) release rpm-4.0.2
>: 	b) release rpm-3.0.7
>: Both are pending, 4.0.2 first.
>   [...]
>: 
>: 73 de Jeff
> 
> ----------------
> From: Jeff Johnson <jbj@JBJ.ORG>
> To: rpm-list@redhat.com
> Subject: rpm-4.0.2 release
> Message-ID: <20010315110245.A4616@eastham.redhat.com>
> Date: Thu, 15 Mar 2001 11:02:45 -0500
> 
>: Rpm 4.0.2 is now available at
>   [...]
>: 
>: A short note on rpm versions:
>: 
>: rpm 3.0.7 is the current "production" version of rpm which is going to be
>: replaced by rpm-4.0.2. There will be a final release of rpm-3.0.7 on
>: ftp.rpm.org, otherwise this is end of life for rpm-3.0.x.
>   [...]
>: 
>: 73 de Jeff
> 
> ----------------
> 
> So is rpm-3.0.7 ever going to be released?  It's been promised for some
> time, and this is all that's on ftp.rpm.org:
> 
>   ftp://ftp.rpm.org/pub/rpm/dist/rpm-3.0.x
>   lftp> dir rpm-3.0.6.tar.gz
>   -rw-rw-r--   1 2369     300       1646191 Sep 13  2000 rpm-3.0.6.tar.gz
>   ftp://ftp.rpm.org/pub/rpm/dist/rpm-3.0.x
>   lftp> dir rpm-3.0.7.tar.gz
>   ftp://ftp.rpm.org/pub/rpm/dist/rpm-3.0.x
>   lftp> dir rpm-3.0.7.tar.bz2
>   ftp://ftp.rpm.org/pub/rpm/dist/rpm-3.0.x
>   lftp> dir *3.0.7*
>   ftp://ftp.rpm.org/pub/rpm/dist/rpm-3.0.x
>   lftp> cd ../..
>   <--- 250 CWD command successful.
>   cd ok, cwd=/pub/rpm
>   ftp://ftp.rpm.org/pub/rpm
>   lftp> dir
>   total 20
>   -rw-rw-r--   1 2370     300           125 May 16  1997 README
>   drwxrwxr-x  11 2370     300          1024 Sep 14  2000 dist/
>   drwxrwxr-x   2 2370     300          1024 Nov 16  1999 docs/
>   drwxrwxr-x   2 2370     300          3072 Dec 22  2000 test/
>   drwxrwxr-x   2 2369     300          1024 Dec 22  2000 test-3.0.7/
>   drwxrwxr-x   2 2369     300          2048 Apr 25 18:03 test-4.0.3/
>   drwxrwxr-x   2 2369     300          1024 May 23 19:49 test-4.0.3-ROLLBACKS/
>   ftp://ftp.rpm.org/pub/rpm
>   lftp> cd test-3.0.7
>   <--- 250 CWD command successful.
>   cd ok, cwd=/pub/rpm/test-3.0.7
>   ftp://ftp.rpm.org/pub/rpm/test-3.0.7
>   lftp> dir
>   total 0
>   ftp://ftp.rpm.org/pub/rpm/test-3.0.7
>   lftp> 
> 
> The release of rpm-3.0.7 has been pending since September of last year,
> about nine months.  I'd guess that no one on this list actually
> believes that it's going to happen, but some sort of announcement that,
> "We lied, we're really not going to release it," would be nice.
> 

Yup, all my words, call it lies if you must.

At this point, I see nothing but confusion from a rpm-3.0.7 release because:

	a) the libio interface has changed incompatibly in glibc-2.2, with some
	fairly complicated debugging on my part for those, like Conectiva,
	that tried to use glibc-2.2 with rpm-3.0.7, as the changes are
	not back ported there.

	b) the bzip2 interface has changed incompatibly, with a hunk of
	changes also necessary to be back ported. Further complications
	come from the fact that Red Hat has not released an errata
	(because none is necessary) to use the bzip2 api on Red Hat 6.2.

	c) the so called "database corruption" qua opportunities for mischief
	which I've spent a whole lot of time fixing in rpm-4.0.3 are all
	still just just waiting to be exercised in rpm-3.0.7. Can you
	say "free list corrupt" ? I knew you could ...

	d) a bunch of weird pokery jiggery in python callbacks to try to
	get a callback for erased packages even though there was no
	way to associate a key (i.e. python object) present at all in
	the rpm API is just as broken as it always is/was. BTW, a 2 month
	delay in getting a reproducible test case necessary to decide
	that the whole mess was crap is why rpm-3.0.7 was not released
	last fall as promised. All this still needs to excised from
	the rpm-4.0.3 API as well.

	e) simultaneous support for database iterators, introduced in
	rpm-4.0, as well as the Good Ol' rpmlib API, in bindings and
	installers and the handful of programs that happen to use rpmlib
	complicates my life immeasurably. All that Good Ol' stuff is
	just as undocumented as it ever was, so I get to do all the
	ports, to all the programs, for no very important purpose.

There's much more, if you wish to hear me rant on. I see nothing but
insanity, much of it not mine, in the sequence of rpm versions

	rpm-3.0.4
	rpm-4.0
	rpm-3.0.5	# released as errata to handle v4 packaging
	rpm-3.0.6	# one line spec file typo with no QA prevents install
	rpm-4.1		# flipped back to rpm-4.0.1
	rpm-4.0.1	# dead on arrival
	rpm-4.0.2	# phew, I finally got an all platform version released
	rpm-4.0.3	# mid July, coincident with db-3.3.x release

and you wish me to continue this madness by releasing rpm-3.0.7 ???

Look, I'm one guy, at the very bottom of the package food chain here at
Red Hat, working 60+ hours a week, and the above is the very best I could
manage in the last year.

You can find rpm-3.0.7 on the rpm-3_0_5 branch in CVS if you want it. I
need to focus on creating a useful, stable, well documented, well maintained
code base, not continue trying to preserve rpm's rich and checkered legacy.

73 de Jeff

-- 
Jeff Johnson	ARS N3NPQ
jbj@jbj.org	(jbj@redhat.com)
Chapel Hill, NC