As of d62ad96, when the connection can't receive more data from the
client, it tells the actor to stop the connection and calls
disable_recv(). The actor operates in it's own thread and when it stops
the connection, disable_recv is being called again from a different
thread. Since the actor is told to stop the connection before
disable_recv is called, the two calls to disable_recv may happen
simultaneously.
This causes a race condition issue where both threads can reach past the
check that recv_id is not None before either of them set it to None. If
one of them set it to None before the other one tries to use it, an
error is raised.
This commit calls disable_recv before telling the actor to stop the
connection. Since disable_recv is a blocking call, this ensures that
recv_id is being set to None before the actor thread begins to stop the
connection.
Fixes#781
KeyboardInterrupts are not always catched in commands.py. As far as I
have seen, if I/O operations are ongoing, a KeyboardInterrupt may be
raised there instead of in commands.py. In some cases, this may cause
loop.run() to return, which previously resulted in exit_status_code
being referenced before assignment.
For the purposes of this event we consider the outputs sink the final element.
If we don't do this we would get one event per branch, and we don't want to
track when each of them actually switches any way. So just tracking when the
tee/outputs bin gets the event is good enough for us.
As part of this I've also added 'testoutput' as a special cased output value.
This is now needed as outputs are always synced to the clock, making testing
a lot less practical.
The default when loading config for logging from a file is to disable
existing loggers. Since some loggers are created before logging is set
up, these loggers were disabled if logging/config_file is set.
(cherry picked from commit cb0387c46d)
Conflicts:
docs/changelog.rst
The default when loading config for logging from a file is to disable
existing loggers. Since some loggers are created before logging is set
up, these loggers were disabled if logging/config_file is set.
Previously, test_list_album_with_artist_name would only test that the
command didn't fail. Now it also checks that the response is correct.
That is, that the response contains albums.
This makes the test detect the error reported in #817.
(cherry picked from commit 9771eda230)
Previously, test_list_album_with_artist_name would only test that the
command didn't fail. Now it also checks that the response is correct.
That is, that the response contains albums.
This makes the test detect the error reported in #817.
This rips the mixer bits and pieces that have been hiding in the audio actor to
it's own class. The software mixer now only knows about this and nothing else
from audio.
This adds an extra mopidy.audio.gst logger and moves the GStreamer logging to
it. Additionally this adds more logging so we can likely get by with just
mopidy logs in more cases.
