bits from the ci.debian.net maintainers
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256
Dear all,
With the recent upload of autopkgtest 5.15, which is available in
unstable and testing, maintainers have the ability to declare on which
architecture(s) their autopkgtests are supposed to be run [1]. This
can be done in either explicitly listing supported architectures, or
by explicitly listing unsupported architectures. The normal
dpkg-architecture wildcards apply. As an example, src:lxc already made
the change in commit [2].
The reason why this feature has been added is that the testing
migration software that is run by the Release Team is now taking test
results into account from four architectures: amd64, arm64, armhf and
i386. And more architectures to be added soon hopefully.
Of course it's best that your autopkgtest is supported on all
architectures, but if that isn't feasible, too much work, or doesn't
work e.g. because of non-installability of test dependencies [3],
consider using this new field.
Paul Gevers
Member of the debian-ci team
Member of the Release Team
[1]
https://salsa.debian.org/ci-team/autopkgtest/-/blob/master/doc/README.package-tests.rst#L178-187
[2]
https://salsa.debian.org/lxc-team/lxc/-/commit/18904a14a91cf3bd705d4a9c62cc64e8dd464df3
[3] We think it's better to use this new field instead of the
skip-not-installable restriction if the problem is architecture
specific, as skip-not-installable will hide genuine problems with
installability. Regular package dependencies are already checked by
the migration software so shouldn't be an issue.
-----BEGIN PGP SIGNATURE-----
iQEzBAEBCAAdFiEEWLZtSHNr6TsFLeZynFyZ6wW9dQoFAl+bAXUACgkQnFyZ6wW9
dQqP2QgAo4rP/ZVSvCozQ9B8QP5EPbxmHJcwewy4AHv5O+dPEch4Wf8OP6SI4lTg
bqbfQMId4ArIvwzTWNNlrIpa4XSKxG4q9v6kuj8CHNxboI2kYNxju2g0BEoGrGS8
heOenMF2GrrQydBMcztWPYbpnIGyLWDz7xK4Pwl6WnBjwwH8U2oiuN6jbGgjRqIq
3a6Nb9gXAIpY5qiT/ToJTyr00UChFofhfKCeWGr2d8k2GKKy3x7ATPJsn4XT5pNn
Bu+FbKkF9VbrnReh+Hyhbb6hTeaI2bvuB2xx6Rbee3VRNuxHSq4YkD49K9YKZ/YO
MHg2Xv/w/Xcve1Nd9oXjyzohXa5YNg==
=VMG0
-----END PGP SIGNATURE-----
Reply to: