Hi Christoph,
On Mon, Feb 05, 2024 at 02:46:35PM +0100, Christoph Berg wrote:
> Control: tags -1 = moreinfo
> Re: Steve Langasek
> > If you have any concerns about this patch, please reach out ASAP. Although
> > this package will be uploaded to experimental immediately, there will be a
> > period of several days before we begin uploads to unstable; so if
> > information
> > becomes available that your package should not be included in the
> > transition,
> > there is time for us to amend the planned uploads.
> Hi,
> I just found out that libpg-query is included because it was thought
> to be "uninstallable":
> https://adrien.dcln.fr/misc/armhf-time_t/2024-02-01T09:53:00/logs/libpg-query-dev/apt.log
> [2024-01-20T03:02:49+00:00] apt-get install libpg-query-dev libprotobuf-c-dev
> postgresql-server-dev-15 abi-compliance-checker
> E: Unable to locate package postgresql-server-dev-15
> I think that's bogus, the package has not been depending on PG15 for
> some time.
The sequence here is:
2023-03 attempting to analyze libpg-query-dev fails because of undeclared
header dependencies.
https://people.canonical.com/~vorlon/armhf-time_t/logs/libpg-query-dev/base/log.txt
2023-07 post to debian-devel proposing that -dev packages we can't analyze
in a timely fashion be included in the transition to be safe, and that
maintainers can help analyze if they want their package excluded. Package
list attached which includes libpg-query-dev.
https://lists.debian.org/debian-devel/2023/07/msg00232.html
2023-09 quirk added to the a-c-c wrapper script that tries to pull in
missing dependencies of libpq-query-dev identified via static analysis of
packages failing to be analyzed to date, which at the time included
postgresql-server-dev-15.
2023-11 postgresql-15 removed from sid.
2023-12 first full analysis of Debian sid, at which point the quirk for
libpg-query-dev is out of date.
So we have never had a successful analysis of libpg-query-dev, and from our
side do not know that its ABI is not affected by time_t.
> Please exclude it from the NMUs.
> Also, why did I not get a bug for that? I understand that you can't
> look at 1500 packages individually, but checking the 40-something on the
> https://adrien.dcln.fr/misc/armhf-time_t/2024-02-01T09:53:00/summary/results_uninstallable.txt
> list would surely have been possible?
We didn't file bugs about undeclared header deps, because maintainers were
not very receptive to the first few reports we filed (apparently expecting
-dev packages to only ship headers that can be compiled on Debian is too
much) so we decided we were better off just adding quirks manually to not
block on Debian maintainers.
We didn't file bugs about uninstallable packages because from an archive
perspective, trying to avoid a possibly-unnecessary transition for a library
with a hundred reverse-dependencies is a higher priority than trying to
avoid a possibly-unnecessary transition for a library with two
reverse-dependencies[0], regardless of the particular reason we weren't able
to analyze the package.
If you as maintainer want to close this bug report (indicating that no
transition is required) or un-tag it 'pending' (indicating that a transition
may be required but the patch is not ready to upload), and accept any
fallout if it turns out this is incorrect, that will mark it so that we will
not include it in NMUs to unstable.
*I* will not be making either of those state changes to the bug, because I
currently don't have proof that the library's ABI is not affected by time_t.
Thanks,
--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer https://www.debian.org/
slanga...@ubuntu.com vor...@debian.org
[0] https://people.canonical.com/~vorlon/armhf-time_t/sorted-revdep-count
signature.asc
Description: PGP signature
