On Nov 19, 2012 10:27 AM, "Oon-Ee Ng" <ngoonee.talk@xxxxxxxxx> wrote: > > On Mon, Nov 19, 2012 at 3:08 PM, Jan Steffens <jan.steffens@xxxxxxxxx >wrote: > > > On Mon, Nov 19, 2012 at 7:50 AM, Oon-Ee Ng <ngoonee.talk@xxxxxxxxx> wrote: > > > Hi, Pulseaudio from [testing] just updated to 2.99.2-1, now mpd audio > > > output (running mpd as my own user) fails with the following:- > > > > > > Nov 19 14:45 : output: Failed to enable "MPD Pulse Output" [pulse]: > > > pa_context_connect() has failed: Connection refused > > > Nov 19 14:45 : player_thread: problems opening audio device while playing > > > "PointBreak-StandTough.mp3" > > > > > > Setting up pulseaudio to listen on TCP locally as in the wiki works, but > > I > > > never needed to do that before. Can anyone else verify this? Is it > > perhaps > > > logind related (worked fine before this last pulseaudio update)? > > > > You're running mpd from systemd? I believe it's because PulseAudio now > > uses $XDG_RUNTIME_DIR/pulse instead of /tmp/pulse-XXXXXX, > > $XDG_CONFIG_HOME/pulse instead of $HOME/.pulse, and > > $XDG_CONFIG_HOME/pulse/cookie instead of $HOME/.pulse-cookie. > > > > mpd has no session, so these variables are not set, unlike the session > > you login to. > > > > This should be sent upstream (bugs.freedesktop.org). > > > > Hi Jan, thanks for the reply. Sorry for the slightly clueless follow-up, > but it's unclear to me whether 'upstream' here refers to pulseaudio, > systemd, or mpd. Or even whether its actually wrong behaviour. > > My understanding of your explanation is that there's no clear > responsibility. mpd does not use (or require) a session, the locations > pulse uses are fairly standard (I've always hated apps using ~/.foo)... This was recently discussed: http://arunraghavan.net/2012/11/pulseconf-2012-report/ Does not look like anything has come of it yet. I guess it might make sense to ask advice from the PA guys though, they would hopefully have an idea. Tom
