[vlc-devel] Flamebait: PulseAudio removal

salsaman salsaman at gmail.com
Sun Dec 5 01:35:48 CET 2010

I have implemented PA in LiVES for both playback and recording. I can
add the follwoing points:
1) once I had the client coded correctly, I have not observed any
memory leaks due to pulse.

2) pulse implementation is way easier than jack, because jack
callbacks forbid any kind of file IO which requires use of ring
buffers and extra threading. Pulse actually works better if you do
file IO in the callback because it uses variable sized buckets. If
your callback takes too long in jack, your client can be kicked from
the jack server. Pulse is a lot more forgiving.

3) A lot of the earlier problems with pulse and ALSA/OSS turn out to
be distro problems. More recent versions of ubuntu for example seem to
have pretty much fixed this.

4) I'm not using gstreamer so anything to do with pulse/gst does not
really affect me. If I was running a more pro-audio setup I would be
using jack with real time threading enabled, but pulse is fine for the
average desktop user.

Conclusion - if you can fix pulse support rather than remove it, a lot
more people will be happy.



2010/12/4 Rémi Denis-Courmont <remi at remlab.net>:
>   Hello,
> On Saturday 04 December 2010, Colin Guthrie wrote:
>> To be fair there was an explanation in the commit log:
> No, there was not.
>> commit fa898f1beef7fae7c58fc1284c50a4cb886ad109
>> Author: Mark Kretschmann <kretschmann at kde.org>
>> Date:   Fri Nov 12 07:03:46 2010 +0100
>>     Revert "Disable usage of Xlib"
>>     This reverts commit 3b1835d70aa38e4d86bcb9024c365e6b353766e4.
>>     This revert was recommended by j-b, as the commit broke Phonon-VLC
>> in many cases.
>>     (mine didn't work at all).
> This does not explain why it would have allegedly "broken" Phonon-VLC. And it
> does not explain how it will not reintroduce the problem that the commit did
> fix.
> Besides, anything that this commit might have "broken" was surely already
> broken to begin with.
>> Although I will grant you that the explanation lacked a technical detail
>> that it probably deserved.
>> I'd be interested in seeing if it can be reinstated sensibly so will try
>> and take a look.
> If that is the quality standard of Phonon development operational mode, I
> think I'd rather not care about Phonon at all and pretend it does not exist.
> --
> Rémi Denis-Courmont
> http://www.remlab.net/
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> http://mailman.videolan.org/listinfo/vlc-devel

More information about the vlc-devel mailing list