Debian Bug report logs - #794194
britney: Strip off Multi-Arch qualifiers in reverse dependency calculation

Reply or subscribe to this bug.

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Martin Pitt <mpitt@debian.org>

To: Debian BTS Submit <submit@bugs.debian.org>

Subject: britney: Strip off Multi-Arch qualifiers in reverse dependency calculation

Date: Fri, 31 Jul 2015 09:06:59 +0200

[Message part 1 (text/plain, inline)]

Package: release.debian.org
Severity: normal
Tags: patch
User: release.debian.org@packages.debian.org
Usertags: britney

Hello,

in Ubuntu's britney instance (which triggers autopkgtests for gating
unstable → testing promotion) we noticed that some reverse
dependencies were not being triggered, for example:

   Package: ipython3
   Depends: python3:any (>= 3.3.2-2~)

(Dep list trimmed). This is because britney's reverse dependency
calculation tries to look up "python3:any" in the package map, which
isn't a valid package name. Thus these multi-arch qualifiers should be
stripped off in register_reverses().

Attached patch does that.

Thanks for considering,

Martin
-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)

[0001-Consider-packages-with-M-A-qualifiers-in-reverse-dep.patch (text/x-diff, attachment)]

[signature.asc (application/pgp-signature, inline)]

Message #10 received at 794194@bugs.debian.org (full text, mbox, reply):

From: Niels Thykier <niels@thykier.net>

To: 794194@bugs.debian.org, Martin Pitt <mpitt@debian.org>

Subject: Re: britney: Strip off Multi-Arch qualifiers in reverse dependency calculation

Date: Sat, 16 Jan 2016 14:34:14 +0000

[Message part 1 (text/plain, inline)]

Control: tags -1 moreinfo

On Fri, 31 Jul 2015 09:06:59 +0200 Martin Pitt <mpitt@debian.org> wrote:
> Package: release.debian.org
> Severity: normal
> Tags: patch
> User: release.debian.org@packages.debian.org
> Usertags: britney
> 
> Hello,
> 
> in Ubuntu's britney instance (which triggers autopkgtests for gating
> unstable →†’ testing promotion) we noticed that some reverse
> dependencies were not being triggered, for example:
> 
>    Package: ipython3
>    Depends: python3:any (>= 3.3.2-2~)
> 
> (Dep list trimmed). This is because britney's reverse dependency
> calculation tries to look up "python3:any" in the package map, which
> isn't a valid package name. Thus these multi-arch qualifiers should be
> stripped off in register_reverses().
> 
> Attached patch does that.
> 
> Thanks for considering,
> 
> Martin
> -- 
> Martin Pitt                        | http://www.piware.de
> Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)

Hi Martin,

Thanks for the patch.

Does this issue still apply to Britney at the current master branch?  We
have removed register_reverses in favour of a different approach for
unrelated reasons.

Thanks,
~Niels

[signature.asc (application/pgp-signature, attachment)]

Message #17 received at 794194-done@bugs.debian.org (full text, mbox, reply):

From: Martin Pitt <mpitt@debian.org>

To: Niels Thykier <niels@thykier.net>

Cc: 794194-done@bugs.debian.org

Subject: Re: britney: Strip off Multi-Arch qualifiers in reverse dependency calculation

Date: Tue, 9 Feb 2016 22:33:05 +0100

[Message part 1 (text/plain, inline)]

Hey Niels,

Niels Thykier [2016-01-16 14:34 +0000]:
> Does this issue still apply to Britney at the current master branch?  We
> have removed register_reverses in favour of a different approach for
> unrelated reasons.

Sorry for the late answer. Indeed that whole code is gone, so this
patch doesn't apply any more. I can't say yet whether this also
applies to the new installability module -- I just saw that this
replaced the reverse dependency trees. I'm using that in my
autopkgtest module to trigger tests for reverse dependencies, so on
the next merge I need to port that. I guess it's not too hard to do, I
just need to dive into the code a bit.

I'll raise this again if this still is an issue with the new code,
until then we can close this.

Thanks!

Martin

-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)

[signature.asc (application/pgp-signature, inline)]

Message #28 received at 794194@bugs.debian.org (full text, mbox, reply):

From: Martin Pitt <mpitt@debian.org>

To: Niels Thykier <niels@thykier.net>, 794194@bugs.debian.org

Subject: Re: Bug#794194: britney: Strip off Multi-Arch qualifiers in reverse dependency calculation

Date: Tue, 12 Jul 2016 09:39:46 +0200

[Message part 1 (text/plain, inline)]

Control: tag -1 confirmed -moreinfo

Hello Niels,

Niels Thykier [2016-01-16 14:34 +0000]:
> Does this issue still apply to Britney at the current master branch?  We
> have removed register_reverses in favour of a different approach for
> unrelated reasons.

Yes, it does still apply. I rebased Ubuntu's britney onto current
master, and the test_multiarch_dep() test fails. So I reopened the
bug.

Current britney uses  apt_pkg.parse_depends() which already has an
option to strip off multi-arch qualifiers. It just needs to be
enabled.

Patch attached.

Thanks,

Martin

[1] https://git.launchpad.net/~ubuntu-release/+git/britney2-ubuntu/tree/tests/test_autopkgtest.py#n1233


-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)

[0001-Consider-packages-with-M-A-qualifiers-for-reverse-de.patch (text/x-diff, attachment)]

[signature.asc (application/pgp-signature, inline)]

Message #37 received at 794194@bugs.debian.org (full text, mbox, reply):

From: Niels Thykier <niels@thykier.net>

To: Martin Pitt <mpitt@debian.org>, 794194@bugs.debian.org

Subject: Re: Bug#794194: britney: Strip off Multi-Arch qualifiers in reverse dependency calculation

Date: Tue, 12 Jul 2016 16:51:00 +0000

Control: tags -1 moreinfo

Martin Pitt:
> Control: tag -1 confirmed -moreinfo
> 
> Hello Niels,
> 
> Niels Thykier [2016-01-16 14:34 +0000]:
>> Does this issue still apply to Britney at the current master branch?  We
>> have removed register_reverses in favour of a different approach for
>> unrelated reasons.
> 
> Yes, it does still apply. I rebased Ubuntu's britney onto current
> master, and the test_multiarch_dep() test fails. So I reopened the
> bug.
> 

Ok thanks. :)

> Current britney uses  apt_pkg.parse_depends() which already has an
> option to strip off multi-arch qualifiers. It just needs to be
> enabled.
> 
> Patch attached.
> 
> Thanks,
> 
> Martin
> 
> [1] https://git.launchpad.net/~ubuntu-release/+git/britney2-ubuntu/tree/tests/test_autopkgtest.py#n1233
> 
> 

We need to handle the multi-arch dependencies explicitly as "pkg:amd64"
is *not* acceptable for an i386 packages (etc.).  Unfortunately, the
proposed patch causes exactly that behaviour[1]. :-/

[1] We have a very crude test for this in
https://anonscm.debian.org/cgit/collab-maint/britney2-tests.git/tree/t/multi-arch-depends/

It probably only covers the most straight forward case.

Message #44 received at 794194@bugs.debian.org (full text, mbox, reply):

From: Paul Gevers <elbrus@debian.org>

To: Debian Bug Tracking System <794194@bugs.debian.org>

Subject: release.debian.org: Update Ubuntu patch

Date: Tue, 20 Dec 2016 11:46:40 +0100

[Message part 1 (text/plain, inline)]

Package: release.debian.org
Followup-For: Bug #794194
User: release.debian.org@packages.debian.org
Usertags: britney
Control: tags 794194 -moreinfo

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

While diving into debci/britney integration, I noticed that the Ubuntu patch to
fix this issue isn't up-to-date anymore in this bug. It seems that the current
implementation in Ubuntu is answering the concerns.

Paul
P.s. I may try to commit the fix myself to the GitHub archive, but I try
to focus a little bit, so I thought to at least let this bug know.

- -- System Information:
Debian Release: stretch/sid
  APT prefers testing-debug
  APT policy: (500, 'testing-debug'), (500, 'testing')
Architecture: amd64 (x86_64)

Kernel: Linux 4.8.0-2-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)

-----BEGIN PGP SIGNATURE-----

iQEzBAEBCAAdFiEEWLZtSHNr6TsFLeZynFyZ6wW9dQoFAlhZDAgACgkQnFyZ6wW9
dQoI4Af+OCo5a5Bs9NEsj5gW/in6W9rub1cGDeP5yCxw3vXdX4R8xzbVAEXi4r6j
fYZOtoOdCU4Q+Mtp+AYwKqeG3cqtAwC9jhSje/8/5tAMbDCDl2e/EJvoJAA78udJ
b9b0TJI0SEk1vuYf1AXDKpednm4U1lhR+6N1JhEUZylsgbgv6ppPJpZr1Ufq8CvP
sI4wKpXBOAx5vq7/N+PqgOA1Is6/a9/1dc+5b206gH1UpU9t7Rp/KqpAbu++4RT2
YRKGzyTx0+IoHQtBNKALt+nJjoUNOo7XcUHeZlaV8CV1l3f1dBGN+n16iU8svcQE
WIFByifN0uvgRQrNoBdDDf7Vfj/eIw==
=AxHY
-----END PGP SIGNATURE-----

[0001-Consider-packages-with-M-A-qualifiers-for-reverse-de.patch (text/x-diff, attachment)]

Send a report that this bug log contains spam.