- use si-audio-adapter, to keep the original test's spirit
of linking two terminal nodes, not two intermediate ones
- setup null sink to be a virtual device sink with 2 channels
and the audiotestsrc to be a stream, so that it is also configured
to have 2 channels via si-standard-link's logic
- don't use a second client core, as it introduces races;
we don't need that anymore because we don't export endpoint-links
With a hacky workaround, because the generate_gir() implementation
has a bug and does not pass the file dependencies properly down
to the custom_target() that it creates internally
- use code blocks properly
- use note/important blocks
- stop using ` for emphasis, use `` or * appropriately
- update info on contributing, community, running, logging
Let g-ir-scanner parse the original headers for the C signatures
and just add the gtk-doc comments in the generated wp-gtkdoc.h file
Also cleanup a bit gen-api-gtkdoc.py and make it omit printing most of the
whitespace clutter that it used to print inside the gtk-doc comments
* Use custom_target() instead of configured shell scripts
* Do not copy all the .rst files in the build directory
* Setup dependencies between targets
* Tidy up dependencies lookup
* Remove unused files
* Upgrade Doxyfile to the latest version and cleanup used options
invalidation may destroy the closure and this will trigger
_wplua_closure_finalize(), which will remove the closure from the
array while the original code is also trying to do the same
if we destroy the closure after it has been removed from the
array, _wplua_closure_finalize() will not remove anything and
will continue happily
We have ended up not using them, so let's not carry them
in the ABI of 0.4
We can always revert that, but let's first decide how
these objects should be used
Sometimes the default device node might not exist when reevaluating endpoints
for the first time on startup, so the policy would link endpoints to another
device node. Then, the default device node appears and the policy moves the
endpoints to the default device node while the previous link has not finish its
activation yet. This race condition can cause endpoint links to fail when being
activated. Delaying the reevaluation of endpoint links until the first client
link is created avoids this issue.
