The Ubuntu image needs a rebuild, because there's already an image with that
same version which lacks meson. And likewise Fedora needs a rebuild, because
we still need python3-pip for two sub-images and this is (probably) the least
bad way to deal with that.
Signed-off-by: Niklāvs Koļesņikovs <89q1r14hd@relay.firefox.com>
According to gkiagia Fedora 37 is a better choice, because it's newer and its
compatibility with Coverity has been verified by WirePlumber's CI setup already.
Therefore upgrading the CI image to F37 should be safe.
Also fixed a typo in a previous commit's comment introduced by autocompletion.
Signed-off-by: Niklāvs Koļesņikovs <89q1r14hd@relay.firefox.com>
Back when it was added, the latest stable version was used. So it makes sense
to keep using using the latest stable image, which is now 3.17.
Signed-off-by: Niklāvs Koļesņikovs <89q1r14hd@relay.firefox.com>
meson prints the following warning:
WARNING: Running the setup command as `meson [options]` instead
of `meson setup [options]` is ambiguous and deprecated.
SOFA is a file format used for storing and accessing spatial audio data, namely head-related transfer functions. These can be used to create binaural spatial sound using head- or earphones.
This commit introduces libmysofa as an optional dependency for loading SOFA files and creates a spatializer plugin for the filter-chain
ci: install libmysofa-devel for full build
ci: bump FDO_DISTRIBUTION_TAG
...except in the build_session_managers job.
This decouples pipewire's CI from wireplumber's dependencies
and potential failures. The build_session_managers job is supposed
to catch session manager integration errors, not any other job.
This should catch any potential issues with conflicting options, missing
libraries, etc.
We need to disable aptx and roc because Fedora doesn't ship those
libraries, and we disable libcamera because it's a moving target and
shouldn't hold up the pipeline.
We allow those jobs to fail so a theoretical issue in the session
managers doesn't prevent our pipeline from completing.
The failure could however be due to our integration of the session
managers, so failures should still be inspected nonetheless.
Coverity scans the source code for static analysis, but its parser
has a couple of flaws...
The most common one is the failure to parse deprecation macros in
enum definitions, like:
```
G_PARAM_PRIVATE GLIB_DEPRECATED_ENUMERATOR_IN_2_26 = G_PARAM_STATIC_NAME,
^
SD_JOURNAL_SYSTEM_ONLY _sd_deprecated_ = SD_JOURNAL_SYSTEM /* old name */
^
```
but also recent gcc built-in macros, like:
```
#if __has_builtin(__builtin_is_constant_evaluated)
^
#if _GLIBCXX_HAS_BUILTIN(__has_unique_object_representations)
^
```
The solution is to replace the code with regular expressions at an earlier
stage. This can be configured directly in coverity with cov-configure
xmltoman looks dead and uses XML as well as pulls in a long list of
Perl dependencies. This replaces it with rst2man that has almost no
dependencies beyond Python, which is already required for the Meson
build system.
Naturally rst2man uses reStructuredText and the pages were rewritten
via regex and manual editing to be as close to original XML format
as possible. A few fixes and updates were done too. Most notably a
note was added to indicate that production deployments should not be
starting the session manager via pipewire.conf file.
There's three slight formatting issues/differences:
1. rst2man produces simpler footer.
2. "-f | --foo=value" confuses the parser and it fails to correctly
add argument specific syntax highlighting to assignment value.
3. XML version had inconsistent use of <arg> and <opt> which has
been partially addressed. But different manual pages still have
their differences to what and how is highlighted.
To avoid adding dependencies that may be considered a bit too recent, let's
build on the current Ubuntu LTS.
Unlike the current Fedora job, this is just the default build with no special
options. They can be added by anyone motivated enough to track down the
required package names :)
Mostly uses the existing infrastructure, but the webrtc canceller has a
fixed blocksize, so we:
1. Use the canceller blocksize if configured
2. Accumulate output data in a ringbuffer
3. Push out the data in the required chunk size
Let's use a template for the generic case of "building on Fedora"
and a named job for the specific job itself.
The real change here is the artifact handling: by default the artifacts
collected for a .build job are only the meson build logs.
The build_on_fedora job collects the installed files as well (to be used in
the pages job later).
Regression introduced in 7e741ef983, the
change of BUILD_ID in that commit caused the pages job to break.
Fix it by relying on the same path expansion we use for the artifacts in the
parent job instead of a hardcoded path.
Simple jobs that run a standard meson build process, once with no options
selected, once with a matrix of various options explicitly enabled or
disabled. This should pick up any accidental build errors.
The CI_JOB_NAME now includes brackets which cause issues in file paths, so
we use the CI_JOB_ID instead as build dir.
Niklāvs Koļesņikovs
Barnabás Pőcze
Dmitry Sharshakov
Jonas Holmberg
George Kiagiadakis
Tycho Haemers
Frédéric Danis
Pauli Virtanen
Bart Ribbers
Peter Hutterer
Nils Tonnätt
Arun Raghavan