Send server change events to subscribers, when default nodes change due
to some nodes being added/removed/changed, also if there is no metadata
change. Make sure the event is consistent with the result from
find_device at the time.
Previously, these were not necessarily sent, or information from
find_device immediately after the event might not reflect the change
yet.
When a sample is playing a cached sample and it disconnects we should
wait until all the pending samples are finished otherwise we stop
their playback and get a chopped or broken sample.
See #595
Assume -EACCES is a normal thing when checking /proc/<pid>/root and
continue with just an info message. Log a warning and assume
flatpak for all other errors.
Fixes#432
Allow only one memblock in the output queue. Otherwise, if the
client is for some reason not reading from the socket, we would queue
an unlimited amount of messages until we get killed by OOM.
Change some warn to debug in the process function for things we
handle from the non-rt thread.
Fixes#488
Pulseaudio unsets the preferred sink/source when stream is moved to a
default target. Emulate this behavior by setting + unsetting
target.node, which informs policy-node to move it accordingly.
PulseAudio starts with underrun_for equal to -1, sets it to 0 when
something was played and increments it from then on with the number
of bytes played.
The message structures returned by pw_protocol_native_connection_get_next
point to data that is contained in the buffer of the connection.
The data was invalidated when pw_protocol_native_connection_get_next was
called the next time, which made the connection loop non-reentrant, in
cases where it was re-entered from demarshal callbacks.
Fix this by allocating new buffers when reentering and stashing the old
buffers onto a stack. The returned message structure is also stored on
the stack to make lifetimes to match.
Use the rate match io are to read the exact amount of samples we
need for this cycle from the ringbuffer. We can then request lower
latency.
There will also be less buffering in the audioconverter, which makes
our timestamps more accurate.
Parse the module arguments according to the pulseaudio syntax,
not our own syntax (which requires "" around a , delimter). Fixes
audio.position=FL,FR argument.
See #527
Make a new complete event and use it instead of the start event
Use the start event at the start of the cycle
Make the profiler also log incomplete graph cycles
Remove the separate error event and merge with the done event with
error code. This then also marks the sample as done when in error
so that it gets cleaned up properly.
Fixes#486
We can't write more to the ringbuffer than its maxsize. Some clients
send more than the negotiated maxsize and cause a crash if we don't
clamp.
See #440
