Use target-object=<serial/name> instead of path=<id> for specifying
sink/src targets. Deprecate path= argument.
Change device provider to preferably expose serials instead of ids.
In a dynamic builder we can only save the offsets in the array and
deref after we completed building everything.
Increase the control limits.
See #2179
Resizes the buffer dynamically. Be careful with getting the address
of a pod in the buffer, it might not be valid after building more stuff
with the builder.
Add an option to do a hilbert transform on the generated rear channels
to do a 90 degree pahse shift on them. This can improve spacialization
of the rear channels.
See #861
Instead of using snprintf to clip the node line to the terminal width,
causing multibyte characters to be split improperly, this lets curses
wrap the text as it normally would, and then overwrites the wrapped
text with the next line, simulating clipping.
pipewire-media-session purposefully makes one of its cores to lag the
other, and then uses it to bind ids it obtained from the faster core.
This no longer works with the registry generation number checks.
It's possible to fix in p-m-s, but we can also add a specific workaround
for it.
This workaround is supposed to be eventually removed. Workarounds for
other apps should not be added.
Not all clients have an existing registry, and the registry generation
number will not be updated for them. However, we would like to check
for stale globals also elsewhere, eg. metadata, and it must work also
in this case.
To avoid failing to update client registry generation, on global
addition which the client would see if it had a registry, send done
message for the new global id instead.
Message footer should be handled before attempting to find the object
the main message is sent to / checking permissions, because it is not
aimed at a specific object. E.g. the registry generation updates should
be handled regardless of whether the main message is valid or not,
because the updates will not be re-sent.
Fixes registry generation updates sometimes going missing.
We now translate device.description to node.description when parsing
the properties so check for node.description and generate one when
it's not available.
Fixes#2166
Use invoke to set the zero-denormals flag from the data thread when
explicitly set. The flag is per thread and should really only be set on
the data thread and only when explicitly enabled.
Fixes#2160
We need to create fake channels when parsing the IEC958 format or
else we get an invalid format and IEC958 passthrough doesn't work.
Ignore the IEC958 formats when collecting formats for the device or else
the fake channels mess with the real channels of the device.
See #1442
PulseAudio will convert the samplespec/channelmap to a format_info
and omits the input format/rate/channels when the fix flags are set. It
does not use the input values at all.
Do the same in pipewire-pulse, make a single format description without
the requested fields.
This also needs a session manager fix to make it deal with those missing
fields.
See #876
Wait the reply to be sent and the sample to complete playback before
freeing the sample. Otherwise it might have been possible that the
sample completes before the reply can be sent.
