headers
Lennart Poettering | 1 Sep 2009 01:17

Re: Priorities for sinks/source

On Sun, 30.08.09 17:31, Colin Guthrie (gmane <at> colin.guthr.ie) wrote:

Heya!

> Do you have any objection to making these priorities user adjustable via  
> the protocol in some way? I know you intend this to be more or less  
> automatic for general out of the box use, but as we've discussed before,  
> the KDE GUI exposes the priority to the user.

Hmpf. Read access to this variable might be an ok thing to do. But
write access? naah

You should not overestimate the meaning of this field. It should be
used only for tie breaking if otherwise more than one sink
(resp. source) are candidates for *a particular purpose*. That means
that most of the time the field is not used and should not be used at
all. 

Now, more specifically, when *should* it be used?

Right now it is used only for selecting the initial default
sink/source. But later on it should be used as well for example when a
sink is removed that was the configured default for a specific role
and we need to find a new sink for a stream of this role. 

As you know I think that the KDE UI for device configuration is not a
particular ... let's say ... convincing design. So trying to bend the
stuff we already have in PA so that it fits what this (misdesigned in
my eyes) KDE UI needs seems the wrong way to do this.
(Continue reading)

Permalink | Reply | 1 comment |
headers
Lennart Poettering | 1 Sep 2009 01:26

Re: PulseAudio on MacOSX

On Sun, 30.08.09 03:21, Jason A. Donenfeld (Jason <at> zx2c4.com) wrote:

> Hello,
> 
> I'm wondering the status of PA on OSX. The latest SSE commits cause the
> compile to fail, but checking out a revision before those compiles fine.
> Unfortunately, it segfaults when running. What's the timeline for OSX
> support in PA?

There is no timeline. 

Every now and then in the past people staretd to work on porting PA to
MacOSX but unfortunately nobody really pushed all the way through with
with this.

Some patches have been merged that allowed PA to compile on MacOS at
those times. But there never was an output module so it probably was
not useful for much.

> (which crash my router)? Or what about using PA's raop sink and making OSX
> emulate airtunes? Unfortunately, PA doesn't seem to support Airfoil. Links
> and whatnot are in this blog entry:
> http://blog.jasondonenfeld.com/188 .

I have no idea what airfoil is. If you use PA's raop support it will
allow PA to contact an Airport and make its audio device available for
local clients. PA in this way acts as iTunes compatible client.

The other way round, i.e. making PA act as Airport audio device cannot
be implemented right now, since Apple is an evil company and uses
(Continue reading)

Permalink | Reply | 8 comments |
headers
Nasa | 1 Sep 2009 02:07
Picon

Re: Sending all audio out S/PDIF


----- "Lennart Poettering" <lennart <at> poettering.net> wrote:

> On Sat, 29.08.09 17:40, Nasa (nasa01 <at> comcast.net) wrote:
> 
> > Maybe I didn't quite explain it right...  If I login into a terminal
> and
> > do a pulseaudio --start, then a "mplayer *mp3" I will get the song
> playing..
> > If I then do a startx and from the frontend open up a new terminal
> and then
> > do a "mplayer *mp3" I will get connection refused errors.  "whoami"
> in both 
> > terminals will return the same user.
> > 
> > So call me confused...
> 
> Try "PULSE_LOG=99 pactl stat". It will then show you where it tries
> to
> connect.
> 
Back again...

I finally got some sound to work.  I can start playing some music, but after 2 - 10
minutes sound stops.  Depending on the player it may freeze or it may continue with 
out noticing sound has stopped...

Here's what gets logged in my syslog file (it can be repeated a couple of times):

When pulseaudio crashes:
(Continue reading)

Permalink | Reply | 1 comment |
headers
Lennart Poettering | 1 Sep 2009 03:45

Re: Sending all audio out S/PDIF

On Tue, 01.09.09 00:07, Nasa (nasa01 <at> comcast.net) wrote:

> I finally got some sound to work.  I can start playing some music, but after 2 - 10
> minutes sound stops.  Depending on the player it may freeze or it may continue with 
> out noticing sound has stopped...
> 
> Here's what gets logged in my syslog file (it can be repeated a couple of times):
> 
> When pulseaudio crashes:
> Aug 31 04:25:43 ice-car pulseaudio[3109]: rtpoll.c: Assertion 'usec <= ((pa_usec_t)
1000000ULL)*60ULL*60ULL' failed at pulsecore/rtpoll.c:516, function
pa_rtpoll_set_timer_relative(). Aborting.
> Aug 31 04:34:31 ice-car pulseaudio[7316]: pid.c: Stale PID file, overwriting.

This is almost certainly triggered by a bug in your audio driver which
overflows in snd_pcm_delay()/snd_pcm_avail() which causes PA's time
calculations to overflow as a result which is then caught by this
assert. PA >= 0.9.15 is actually very explicit about this issue and
tells you in syslog the driver that is broken and how exactly the
function overflowed.

Many of these unreliabilities of the timing primitives have been fixed
in the drivers in the recent passed. Make sure you run a fully patched
recent kernel.

And of course, in Fedora we tend to ship a kernel that has these
issues mostly fixed.

Lennart
(Continue reading)

Permalink | Reply | 0 comments |
headers
Colin Guthrie | 1 Sep 2009 11:25
Picon

Re: Priorities for sinks/source

'Twas brillig, and Lennart Poettering at 01/09/09 00:17 did gyre and gimble:
>> I intend to add this priority capability into the module-history that  
>> I'm working on (which remembers past sinks and sources and as a side  
>> effect allows sinks/sources to be renamed).
> 
> Ah, heh. I wrote that parapgraph above before I read this one of
> yours... ;-)

:D FWIW, I got my own name wrong! I actually called it 
module-device-manager. That certainly doesn't help things :p

URL is here in my "history" branch (hence why I called the module the 
wrong thing!):
http://colin.guthr.ie/git/pulseaudio/log/?h=history
(freshly rebased but not compile tested, so may be busted...).

I also have patches for pavucontrol to do the renaming (or redescribing 
more accurately) itself.

> Could you elaborate in more detail what this module is supposed to do
> and how it relates to m-s-r?

Probably covered here:
http://colin.guthr.ie/2009/06/whats-cooking-in-the-pulse-pot/

> As I understood you by "renaming" you are referring to the device
> description, not the sink/source identifying name, right?

Yeah, changing the description. I needed a way to query pulse for a list 
of sinks, both currently available and previously active but no longer
(Continue reading)

Permalink | Reply | 0 comments |
headers
Nasa | 1 Sep 2009 13:00
Picon

Re: Sending all audio out S/PDIF


----- "Lennart Poettering" <lennart <at> poettering.net> wrote:

> On Tue, 01.09.09 00:07, Nasa (nasa01 <at> comcast.net) wrote:
> 
> > I finally got some sound to work.  I can start playing some music,
> but after 2 - 10
> > minutes sound stops.  Depending on the player it may freeze or it
> may continue with 
> > out noticing sound has stopped...
> > 
> > Here's what gets logged in my syslog file (it can be repeated a
> couple of times):
> > 
> > When pulseaudio crashes:
> > Aug 31 04:25:43 ice-car pulseaudio[3109]: rtpoll.c: Assertion 'usec
> <= ((pa_usec_t) 1000000ULL)*60ULL*60ULL' failed at
> pulsecore/rtpoll.c:516, function pa_rtpoll_set_timer_relative().
> Aborting.
> > Aug 31 04:34:31 ice-car pulseaudio[7316]: pid.c: Stale PID file,
> overwriting.
> 
> This is almost certainly triggered by a bug in your audio driver
> which
> overflows in snd_pcm_delay()/snd_pcm_avail() which causes PA's time
> calculations to overflow as a result which is then caught by this
> assert. PA >= 0.9.15 is actually very explicit about this issue and
> tells you in syslog the driver that is broken and how exactly the
> function overflowed.
>
(Continue reading)

Permalink | Reply | 3 comments |
headers
Colin Guthrie | 1 Sep 2009 13:35
Picon

Re: Sending all audio out S/PDIF

'Twas brillig, and Nasa at 01/09/09 12:00 did gyre and gimble:
> 1. add option snd-intel8x0 ac97_quirk=3 to alsa-base in /etc/modprobe.
> 2. add tsched=0 option to module-alsa-sink
> 
> Do you know if these are good suggestions and/or if there are better ones?  

Passing tsched to module-alsa-{sink,source,card} or to 
module-{udev,hal}-detect is a work around and disables glitch-free mode. 
This should not really be a "solution" to the problem which, as lennart 
pointed out is to fix the driver in the kernel.

You should really speak to the Ubuntu kernel people and ensure the 
relevant fixes from Fedora or upstream are included in the kernel.

Col

--

-- 

Colin Guthrie
gmane(at)colin.guthr.ie
http://colin.guthr.ie/

Day Job:
   Tribalogic Limited [http://www.tribalogic.net/]
Open Source:
   Mandriva Linux Contributor [http://www.mandriva.com/]
   PulseAudio Hacker [http://www.pulseaudio.org/]
   Trac Hacker [http://trac.edgewall.org/]
Permalink | Reply | 0 comments |
headers
Lennart Poettering | 1 Sep 2009 18:07

Re: Sending all audio out S/PDIF

On Tue, 01.09.09 11:00, Nasa (nasa01 <at> comcast.net) wrote:

> 1. add option snd-intel8x0 ac97_quirk=3 to alsa-base in
> /etc/modprobe.

Noe idea what the effect of this is.

> 2. add tsched=0 option to module-alsa-sink

This is ugly. And is just a workaround for a broken kernel driver.

> Do you know if these are good suggestions and/or if there are better ones?  

Yes, get a fixed kernel.

Lennart

--

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net
http://0pointer.net/lennart/           GnuPG 0x1A015CC4
Permalink | Reply | 1 comment |
headers
Jason A. Donenfeld | 1 Sep 2009 19:13
Favicon

Re: PulseAudio on MacOSX

On Mon, Aug 31, 2009 at 7:26 PM, Lennart
Poettering<lennart <at> poettering.net> wrote:

> I have no idea what airfoil is. If you use PA's raop support it will
> allow PA to contact an Airport and make its audio device available for
> local clients. PA in this way acts as iTunes compatible client.

Airfoil is a client and server implementation of the Airtunes 2
protocol, the second generation of RAOP that supports timing
synchronization via UDP. This would be nice to have in PA. Col and I
have been talking about this privately; I'll get back to you.
Permalink | Reply | 7 comments |
headers
Lennart Poettering | 1 Sep 2009 19:26

Re: PulseAudio on MacOSX

On Tue, 01.09.09 13:13, Jason A. Donenfeld (Jason <at> zx2c4.com) wrote:

> 
> On Mon, Aug 31, 2009 at 7:26 PM, Lennart
> Poettering<lennart <at> poettering.net> wrote:
> 
> > I have no idea what airfoil is. If you use PA's raop support it will
> > allow PA to contact an Airport and make its audio device available for
> > local clients. PA in this way acts as iTunes compatible client.
> 
> Airfoil is a client and server implementation of the Airtunes 2
> protocol, the second generation of RAOP that supports timing
> synchronization via UDP. This would be nice to have in PA. Col and I
> have been talking about this privately; I'll get back to you.

Uh, is that "second generation RAOP" something apple cme up with? Or
does only "airfoil" speak this?

Does "airfoil" support emulating an airport? i.e. receiving audio from
iTunes?

Are there any docs available?

Lennart

--

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net
http://0pointer.net/lennart/           GnuPG 0x1A015CC4
(Continue reading)

Permalink | Reply | 6 comments |

Gmane