There is a little bit more on "removing packages" here:
So it's actually a 'must' to have a LP bug for getting a package removed.
BR, Frank
On Sun, May 5, 2024 at 7:32 PM Simon Quigley <simon@tsimonq2.net> wrote:
Hi Dima,
As a Debian Developer myself, I understand your concerns. Processes in
this respect could be slightly better, but it also comes down to the
differences between the two distributions. More detailed responses inline.
On 5/2/24 10:56 AM, Dima Kogan wrote:
> Hello.
>
> I'm a Debian developer, and contribute to Ubuntu only indirectly: by my
> contributions to Debian being automatically pulled into Ubuntu. But
> since Ubuntu has more users than Debian, most of MY users use the Ubuntu
> packages. So I'd like to talk about improving the links between the two
> projects. In particular:
Ubuntu and Debian package maintenance responsibilities are slightly
different; in Ubuntu, members of the Core Developers team are
collectively responsible for the packages in the Main and Restricted
components, and Masters of the Universe are collectively responsible for
packages in the Universe and Multiverse. The ratio of "maintainers
holding responsibility":"packages to be maintained" is much lower in
Ubuntu than it is in Debian.
Once a package has landed in the Ubuntu archive, Ubuntu Developers now
collectively hold responsibility for that package. We ease much of this
work by autosyncing packages without deltas to Ubuntu in the first half
of each cycle; that being said, we sometimes drive major transitions in
Ubuntu before Debian, to align with our release cycle.
Many Ubuntu Developers (myself included) are trained to give as much
back to Debian as we possibly can. If we fix a package that both exists
in Debian and has the same bug, we are encouraged to send that fix
upstream to the Debian bug tracker (or upstream itself, or both) to
ensure less friction when we have to merge new changes from Debian. Some
teams within Ubuntu do not follow this process at all, but I would
consider them the exception rather than the rule.
The Debian maintainers of a package are not responsible for how their
packages are used in Ubuntu, that's Ubuntu's responsibility. That being
said, it is best practice to collaborate as much as we reasonably can,
with the time we are given.
> 1. Debian and Ubuntu both have separate bug trackers. But for most
> Ubuntu packages, there's no "Ubuntu" maintainer: there's just the
> indirect one from Debian. In this case (which is most packages), it's
> unhelpful for the Ubuntu bug tracker to exist as a separate thing. If
> it must exist as a separate thing, those bugs should be forwarded
> automatically to the Debian bug tracker. And updates (including
> status updates) should be ingested back into the Ubuntu tracker. For
> my packages I do try to manually look at the Ubuntu bug reports, but
> I have no rights to close those bugs on launchpad. Probably I can
> sign up somewhere, but as the DEBIAN developer, I shouldn't need to
> do that.
I disagree with this approach. Ubuntu and Debian are not ABI-compatible;
Ubuntu has a slightly different toolchain than Debian, and there are
core differences in e.g. dpkg. Not all Ubuntu bugs are Debian bugs, not
all Ubuntu teams want their bugs sent up to Debian, and many Debian
Maintainers don't care about Ubuntu. This is a reality of maintaining
separate distributions.
In some common cases, yes this seems reasonable, we should forward bugs
to Debian. That being said, the first step is making sure the bug
actually exists in the Debian-built version of the package, which is not
always the case.
Generally, I do think we can be better about triaging our bugs and
sending what we can up to Debian. That being said, I disagree with the
solution of completely automating it.
> 2. As I just discovered, when Ubuntu rebuilds the archive for a release,
> packages that FTBFS are silently dropped. There's no bug report on
> either of the two bug trackers. I'm upstream for a project that has
> been excluded from 24.04 because of this gap in the process. There
> really should be a bug report filed, so that the problem can be fixed
> before the release (what Debian does for their releases). And this
> should be filed on the Debian bug tracker, if that's where the
> maintenance happens.
This entirely falls on the Ubuntu Archive Administrators. To my
understanding as an Ubuntu Developer, if we want a package removed, it
is best practice to either have a Debian removal bug or an Ubuntu
removal bug explaining the rationale. Whether this is enforced is up to
the Archive Administrator doing the removal, since the only known public
documentation says nothing about filing bugs:
https://wiki.ubuntu.com/ArchiveAdministration#Removals
To say that packages that FTBFS are indiscriminately removed during
transitions ignores the fact that usually, we do have to file a bug if
we want an AA to remove a package.
> I don't know how hard the above is, but the current situation isn't
> great for either of us.
>
> Related question: is there any way to get my packages included into some
> sort of noble "updates", or something like that?
>
> I'm looking at the "mrcal" source package that had an ininteresting
> FTBFS bug due to some dependency changing its interface. There was a
> Debian bug report filed and quickly fixed, but this happened too late
> for noble:
>
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1067398
>
> The fix is to update the "mrbuild" package to at least 1.9. Is it
> possible to get an updated "mrbuild" and "mrcal" into 24.04?
>
> If I'm misinterpreting what's going on, please let me know. Right now I
> see this:
>
> dima@shorty:~$ rmadison -u ubuntu mrbuild | grep noble
> mrbuild | 1.8-1 | noble/universe | source, all
>
> dima@shorty:~$ rmadison -u ubuntu libmrcal-dev | grep noble
> libmrcal-dev | 2.4.1-1build1 | noble-proposed/universe | arm64, ppc64el, riscv64
>
> The latest mrcal IS 2.4.1, but here it's in "noble-proposed" and not for
> amd64 for some reason.
I would highly suggest filing a bug against the package in Launchpad
following the Stable Release Update procedure:
https://wiki.ubuntu.com/StableReleaseUpdates#Procedure
If you can articulate your case and the rationale for the update well,
it is unlikely that it will be rejected.
Best regards,
--
Simon Quigley
simon@tsimonq2.net
@tsimonq2:ubuntu.com on Matrix
tsimonq2 on LiberaChat and OFTC
5C7A BEA2 0F86 3045 9CC8
C8B5 E27F 2CF8 458C 2FA4
--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
