Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

The new `aptly` publisher, as a feature, allows multiple versions of
the package to be published in the same repository. Note that the
previous publisher, `reprepro` would publish only the single, latest version.

Signed-off-by: Ritesh Raj Sarraf <ritesh.sarraf@collabora.com>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Signed-off-by: Apertis CI <devel@lists.apertis.org>

Since the pipeline is meant to be `include:`d from other repositories
by default it tries to retrieve the `update_test_binaries.sh` script
with a HTTP call.

However this means that the test pipeline on a specific commit is not
actually testing the `update_test_binaries.sh` script on the same
commit, but it retrieves it from the `$osname/$release` branch (in this
case, `apertis/v2023dev1`).

Tweak things to ensure the current `update_test_binaries.sh` is used
during testing.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

In the past, using the parent directory as the working directory has been
a source of troubles in the CI: since it does not get cleaned up on
runners that process multiple jobs (as the lightweight one does) stale
files from other jobs have been a source of failures.

The previous commits have switched the update script to use `$TMPDIR` as
the working directory anyway, but to be extra sure let's not `cd ..`
before invoking the script.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

The `$TMPDIR` is generally not shared in containers setup, so there's
less chance to involuntarily use cached outdated files.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

If the value of a environment variable contains a newline, a naive
`sort` will mangle the output.

Use null-terminated chunks to handle newlines properly.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Some tools stopped being relevant once we moved from Jenkins to GitLab a
long time ago, let's clean them up.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

When running `update_test_binaries.sh` for two different branches
without cleaning up files in between, the `Packages` indexes get re-used
even if they target the wrong release.

This means that if you update the `v2022pre` binaries, then switch to
`v2021` and do the update, the tool will re-use the `v2022pre` indexes
and retrieve the `v2022pre` packages rather than the `v2021` ones.

By including the release in the locally cached `Packages` files we avoid
the issue.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

After the Bullseye rebase the installed tests in `glib-gio-fs` changed
location, so we need to checkout a commit recent enough that the
`external-binaries.cfg` points to the right paths.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

We've moved to GitLab a while ago.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Use committer name/email that match the one of the actual GitLab user
configured to push commits from the CI/CD pipeline.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

Lightweight CI runners have been introduced to run jobs which have low
resource consumption and do not need privileged containers to reduce costs
and latency, so let's use them.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>

Signed-off-by: Ritesh Raj Sarraf <ritesh.sarraf@collabora.com>

Signed-off-by: Ritesh Raj Sarraf <rrs@debian.org>

The latest addition failed to run. This appears to be as a result of the
lack of a new line at the end of the file.

The latest addition is also not in alphabetical order when the others are
so re-sort the ordering.

Signed-off-by: Martyn Welch <martyn.welch@collabora.com>

Signed-off-by: Baghmar Tarun <tarun.baghmar@in.bosch.com>

Rather than moving the global pipeline as is from Jenkins to GitLab CI,
make use of the include mechanism to run isolated pipelines on each test
repository. This makes things more resilient since a broken repository
does not cause the whole process to fail, and also gives more
flexibility since new test repositories can be set up by anyone with no
need to list them here.

Signed-off-by: Emanuele Aina <emanuele.aina@collabora.com>