vcswatch reports that
this package seems to have a new changelog entry (version
5.44, distribution
UNRELEASED) and new commits
in its VCS. You should consider whether it's time to make
an upload.
Here are the relevant commit messages:
commit 3bb86bec2429b122ee0b9a0b5429fa13457dbb0b
Merge: 51fa86d efa31e7
Author: Simon McVittie <smcv@debian.org>
Date: Tue Jan 21 00:14:45 2025 +0000
Merge branch 'wip/smcv/no-more-synth-deps' into 'master'
Fix misleading "synth_deps"/"synth_depends" variable names
See merge request ci-team/autopkgtest!508
commit efa31e71b2c5ea22670ad4b88cd90b3c0fe204ee
Author: Simon McVittie <smcv@debian.org>
Date: Mon Jan 20 11:36:49 2025 +0000
tests: Make assertions about package_under_test_depends, not just depends
Signed-off-by: Simon McVittie <smcv@debian.org>
commit d6bccee90ea2bf0b5f69e0a3c770f508f71d6f57
Author: Simon McVittie <smcv@debian.org>
Date: Mon Jan 20 11:29:23 2025 +0000
Rename synth_deps to package_under_test_deps
This variable was always called synth_deps, synth_depends or similar, for
"synthesised dependencies", but that name is misleading, in both
directions: not every member of the list is a synthesised dependency,
and not every synthesised dependency is a member of the list. What it
*actually* means appears to be: "dependencies that are a binary package
built by the source package under test". Rename the variables to make
that clearer.
The synthesised dependencies were originally the dependencies that were
generated to replace the special token `@`. When `@builddeps@`
and `@recommends@` were added, the corresponding debug messages indicate
that their expansions are also considered to be synthesised dependencies,
but they were not added to this list.
Meanwhile, binary packages that are listed by name *are* added to this
list if they match the name of a binary package built by the source
package under test (with some extra subtleties for alternatives-groups).
The real functional difference between this list and other dependencies,
in adt_testbed, is that for dependencies in this list, we try harder to
install a real package (not merely a virtual package). This has nothing
to do with whether a dependency was synthesised or explicit, but is
appropriate for the binary packages under test - if we are testing, for
example, notification-daemon from the source package of the same name,
then we want to install that specific binary package, as opposed to
some other package like dunst that merely Provides equivalent
functionality.
No functional change intended: this is just clarifying the effect of
the existing code.
Helps: #1052119, !483
Signed-off-by: Simon McVittie <smcv@debian.org>
commit 2c7840bed56ba117c10e3cf98af89d3313e2d8a6
Author: Simon McVittie <smcv@debian.org>
Date: Mon Jan 20 11:15:30 2025 +0000
testdesc: Clarify where synthesised dependencies came from
No functional change, just clearer debug messages.
Signed-off-by: Simon McVittie <smcv@debian.org>
commit 51fa86dd3c4baf388828ed3a5cff4d69c2318aa6
Author: Simon McVittie <smcv@debian.org>
Date: Mon Jan 20 10:52:17 2025 +0000
Update changelog
commit 8515dbf2cef9249dc3003dd6470559c78cbe27e2
Merge: b1503a6 e7b996f
Author: Simon McVittie <smcv@debian.org>
Date: Sat Jan 18 10:16:30 2025 +0000
Merge branch 'wip/smcv/1078205' into 'master'
virt-podman: Document that full systemd functionality needs CAP_SYS_ADMIN
See merge request ci-team/autopkgtest!396
commit b1503a6f744bc33b89d71dad8f3d8195b7c2b089
Merge: 18bed60 9e86d5a
Author: Simon McVittie <smcv@debian.org>
Date: Sat Jan 18 10:10:17 2025 +0000
Merge branch 'wip/smcv/1093275' into 'master'
tests: By default only exercise timeouts on known-fast architectures
See merge request ci-team/autopkgtest!506
commit 9e86d5a2d9c18064fad57d8dc80f6b07c8cd6107
Author: Simon McVittie <smcv@debian.org>
Date: Fri Jan 17 10:26:13 2025 +0000
tests: By default only exercise timeouts on known-fast architectures
We test timeouts by making one part of a test use sleep(1) to block
for a while, trying to guess a timeout *T* that will be long enough to
accommodate initial test setup (up to that point), setting autopkgtest's
timeout to a little more than *T*, and making sleep(1) sleep for
considerably more than *T*. If all goes well, this means that everything
before the targeted point runs (and we can confirm this by screen-scraping
the log), and then the test times out (and we can check that autopkgtest
responds correctly to that).
However, choosing an appropriate value of *T* is difficult. If we make
*T* too long, successfully running the autopkgtest test suite takes that
much longer, even on an arbitrarily fast machine; this is undesirable
because full coverage implies running the test suite several times
with different backends, which can be extremely time-consuming already
(see #1020699). Conversely, if we make *T* too short, we can hit test
failures on slower machines like riscv64, with the test timing out
before initial setup has finished, meaning autopkgtest will (correctly!)
report the failure differently.
In practice the timeouts we have chosen are all tuned for reasonably
fast developer machines like x86, and the code under test is identical
on any architecture, so skip these tests by default on architectures
that are not known to be fast and reliable.
Developers can set `AUTOPKGTEST_TEST_TIMEOUTS=1` to force these tests
to run even on slower architectures, or `AUTOPKGTEST_TEST_TIMEOUTS=0`
to force them to be skipped even on fast architectures.
Signed-off-by: Simon McVittie <smcv@debian.org>
Closes: #1093275
commit 18bed60326297786e07651557df851fedd25aaad
Author: Simon McVittie <smcv@debian.org>
Date: Fri Jan 17 09:26:28 2025 +0000
Remove obsolete alternative dependency on pep8
pycodestyle has been available since at least Ubuntu 18.04 and Debian 10,
which are both older than our documented minimum host system.
Gbp-Dch: short
Closes: #1093161
commit 49b838c6da2738562b511a8ea6dd1196f65629ad
Author: Simon McVittie <smcv@debian.org>
Date: Sat Dec 21 12:30:17 2024 +0000
d/README.source: Mention some apt features that we cannot use yet
Signed-off-by: Simon McVittie <smcv@debian.org>
commit e7b996fdabd815e6d227953aa4459166192c41d0
Author: Simon McVittie <smcv@debian.org>
Date: Mon Dec 9 10:44:18 2024 +0000
virt-podman: Document that full systemd functionality needs CAP_SYS_ADMIN
See https://bugs.debian.org/1078205 and
https://github.com/containers/podman/discussions/23558 for more details.
Closes: #1078205
Signed-off-by: Simon McVittie <smcv@debian.org>
![[bug history graph]](https://qa.debian.org/data/bts/graphs/a/autopkgtest.png){kind=link}