Phonon's backends to JACK.

Harald Sitter sitter at kde.org
Sun Jun 12 22:39:46 CEST 2011


On Sun, Jun 12, 2011 at 9:55 PM, Mark Constable <markc at renta.net> wrote:
> There seems to be a either/or take on Jack vs Pulse whereas I thought
> the promise of Phonon would mean the inclusion of all backend options,
> including a PA-free low latency near zero xrun Jack option regardless
> of CPU cycles involved... if the end user so chooses.

Seeing as there are three people maintaining a code base of about
80.000 lines, of which one actually does PulseAudio exclusively. I'd
argue that if the user chooses to run a setup outside the usage scope
defined by the primary target audience of Phonon, they would need to
live with shortcomings or spend some time on improving the available
code so that it meets their requirements as well as those of the
outlined target audience.
But from a pure target reach point of view it does not appear to be a
worthwhile investment to spend hours on making a setup scenario work
flawless that is neither addressing needs of the target audience nor
those of the three present developers.

That is not to say that the concern is invalid, in fact I fully
appreciate the severity of it. But with a limited supply of resources
one either needs to hope for someone to have a stroke of motivation to
resolve such issues outside the target scope, or that the resource
situation improves and through that time becomes available to spend on
additional "nice-to-have" features that might be generally
uninteresting to everyone on the team, but important enough to friends
within the community that one would still be willing to spend time on
it.

Now, JACK support is exactly that. JACK support from a latency point
of view is absolutely not within the target scope of Phonon, nor
within that of the target audience of Phonon. To that degree it is one
of those "nice-to-have" features. At the same time it makes a lot of
sense to those that use JACK to have equally good JACK integration as
we have with PulseAudio already. I am decently sure that we all can
agree that connecting our frameworks to PA and PA to JACK and JACK to
ALSA ... is certainly not a solution but at best a workaround.

So, yes, it would be very nice to have decent JACK support (even if it
were not at the same level as PulseAudio support). But it is not a
priority and the time (and since Phonon is community driven, to a
large degree also the motivation) are just not available.

What can *you* do to improve the situation?
a) if you are knowledgeable enough, take a look at the
phonon-gstreamer/phonon-vlc code and try to find a way to get JACK to
work, in both cases this should not be too difficult
b) otherwise try to find someone who is equally interested in having
decent JACK support and get this person to take a look at the code
c) try to get more people on the Phonon (developer) bandwagon to free
up resources so that people have more time to spend on projects
outside the target scope

Until then I shall hope that hooking up PA with JACK and thus making
use of our excellent PA integration on the Phonon side of things
should proof a decent enough workaround.

regards,
Harald


More information about the Phonon-backends mailing list