Bug 668

Summary: mythfrontend-0.21-19.fc10.x86_64 missed pulse audio support
Product: Fedora Reporter: thestonewell <thestonewell>
Component: mythtvAssignee: Jarod Wilson <jarod>
Status: RESOLVED FIXED    
Severity: normal CC: ali, bob
Priority: P1    
Version: 10   
Hardware: All   
OS: GNU/Linux   
namespace:
Attachments: Add Pulseaudio headers to Spec file

Description thestonewell 2009-06-16 22:41:08 CEST
with the latest package you now get:

2009-06-16 22:30:59.919 ERROR: ***Pulse Audio is running!!!!***
2009-06-16 22:30:59.919 ERROR: But MythTV has not been compiled with Pulse Audio disabling support. EXITING!

Seems the latest package was not compiled with the pulse audio headers/libs.
Comment 1 thestonewell 2009-06-16 22:43:00 CEST
that is actually for mythfrontend... I assume though that all packages could be affected.

Cheers
Comment 2 Richard Smith 2009-06-17 14:34:23 CEST
Created attachment 202 [details]
Add Pulseaudio headers to Spec file

After a little investigation, it appears that the configure script quietly disables pulse audio if the appropriate header files are not present on the system.

yum install pulseaudio-libs-devel provides the required headers.

I suggest adding pulseaudio-libs-devel to mythtv.spec as a BuildRequires, as in the attached patch.
Comment 3 J. Ali Harlow 2009-06-17 16:06:53 CEST
(In reply to comment #2)
> I suggest adding pulseaudio-libs-devel to mythtv.spec as a BuildRequires, as in
> the attached patch.

I believe this approach should work if the user also sets an environment variable (EXPERIMENTALLY_ALLOW_PULSE_AUDIO). As such it may be an acceptable solution. A better solution would be to patch Myth such that if pulse audio is running and the environment variable is _not_ set then it sould continue to use pulseaudio's alsa emulation layer (ie., as release 18) and if it is set then it should use the new pulseaudio code.

I suggest this because I've never had any problems with Mythtv and pulse audio on my box (heck, I even use it to pipe sound to my headphones via my laptop). It seems that Fedora simply doesn't suffer from the problems the Myth developers were trying to solve (~2s audio latency with significant variability on top).

Ali.
Comment 4 Jarod Wilson 2009-06-17 16:35:05 CEST
*cough* see bug 651 *cough*

I just neglected to commit and build the fix for F10. Will do so shortly.
Comment 5 thestonewell 2009-06-17 16:41:26 CEST
wow.. quick turnaround.. I missed 651 when searching. So sorry for the noise.

I think I can now wait the day or two till the .20 release.

Thanks for the good work!
Comment 6 J. Ali Harlow 2009-06-17 16:43:14 CEST

*** This bug has been marked as a duplicate of bug 651 ***
Comment 7 Jarod Wilson 2009-06-19 02:42:26 CEST
The fix that worked for F11 apparently doesn't for F10, so I'm reopening this bug.
Comment 8 J. Ali Harlow 2009-06-19 03:22:59 CEST
One datapoint:

# ldd /usr/bin/mythfrontend | fgrep pulse
	libpulse.so.0 => /usr/lib/libpulse.so.0 (0x008ea000)

So the BR would appear to have worked.
Comment 9 thestonewell 2009-06-19 08:49:08 CEST
Would explicitly defining USING_PULSE as compile flag help?
Comment 10 Jarod Wilson 2009-06-20 05:37:09 CEST
(In reply to comment #8)
> One datapoint:
> 
> # ldd /usr/bin/mythfrontend | fgrep pulse
>         libpulse.so.0 => /usr/lib/libpulse.so.0 (0x008ea000)
> 
> So the BR would appear to have worked.

I'm slightly befuddled. ldd output looks more or less the same vs. the 32-bit f11 build:

$ ldd mythfrontend |grep pulse
	libpulse.so.0 => /usr/lib/libpulse.so.0 (0x028ef000)

So why would it be fine on F11 and notsomuch on F10?...


(In reply to comment #9)
> Would explicitly defining USING_PULSE as compile flag help?

Dunno. Didn't have to for F11, but... Time for a local test build, I guess...
Comment 11 Jarod Wilson 2009-06-20 07:21:44 CEST
(In reply to comment #0)
> with the latest package you now get:
> 
> 2009-06-16 22:30:59.919 ERROR: ***Pulse Audio is running!!!!***
> 2009-06-16 22:30:59.919 ERROR: But MythTV has not been compiled with Pulse
> Audio disabling support. EXITING!
> 
> Seems the latest package was not compiled with the pulse audio headers/libs.

So now I'm quite befuddled. I got 0.21-20.fc10 installed on an fc10 box with pulseaudio running, and mythfrontend behaves just fine.
Comment 12 Jarod Wilson 2009-06-20 07:24:11 CEST
$ mythfrontend 
2009-06-20 01:20:03.203 Using runtime prefix = /usr
2009-06-20 01:20:04.049 DPMS is active.
2009-06-20 01:20:04.050 Unable to read configuration file mysql.txt
2009-06-20 01:20:04.050 Empty LocalHostName.
2009-06-20 01:20:04.050 Using localhost value of thanatos.wilsonet.com
2009-06-20 01:20:04.065 New DB connection, total: 1
2009-06-20 01:20:04.080 Connected to database 'mythconverg' at host: localhost
2009-06-20 01:20:04.081 Closing DB connection named 'DBManager0'
2009-06-20 01:20:04.086 Primary screen 0.
2009-06-20 01:20:04.097 Connected to database 'mythconverg' at host: localhost
2009-06-20 01:20:04.099 Using screen 0, 1280x1024 at 0,0
2009-06-20 01:20:04.146 AudioPulseUtil: Suspend Success

$ rpm -qf /usr/bin/mythfrontend
mythtv-frontend-0.21-20.fc10.i386

So not a whole lot else I can think of to try right now.
Comment 13 J. Ali Harlow 2009-06-20 10:32:46 CEST
Worked it out: The check is in libmyth and not in mythfrontend. Once I upgraded that as well, it all works. /me hangs head in shame.
Comment 14 Jarod Wilson 2009-06-21 03:28:28 CEST
Oh, good. Because I was rather confused when everything Just Worked with a full 0.21-20.fc10 install last night. :)
Comment 15 Jarod Wilson 2009-06-22 04:16:02 CEST
*** Bug 676 has been marked as a duplicate of this bug. ***
Comment 16 thestonewell 2009-06-23 08:47:33 CEST
updated from testing repo and can confirm that -20.fc10 fixed the issue for me too.

Thanks!