PulseAudio integration issues

[Conn]

Hello,

Feature Freeze is scheduled for Intrepid in eight days, and several
configuration issues relating to PulseAudio have not yet been addressed. I
feel it is very important to encourage some discussion and collaboration
towards fixing these issues (or at least a discussion as to why we won't fix
them).

Here is a list of the bugs which need immediate attention, from most
important to least, with a short summary:

Bug #198453 [1]: We need to enable the PulseAudio ALSA plugins on
configurations where PulseAudio is installed, otherwise ALSA applications
will fail to pass audio to the PulseAudio server. In simpler terms, this
means that users will experience several audio mixing issues with the
current PulseAudio configuration. There are some applications that are not
compatible with these ALSA plugins (Audacity, old WINE releases, Flash <v10,
the stable RealPlayer/Helix Player release), but they are most certainly a
minority, and almost all of them have fixes in progress or pending.

Bug #192888 [2]. Flash v9 is one application that does not work correctly
with the PulseAudio ALSA plugins. Adobe released a support library to allow
extra sound and secure transaction schemes to be supported in Flash called
"libflashsupport", and this has been adapted to support PulseAudio.
Unfortunately this support library causes Flash to become unstable; closing
and opening pages with Flash content (a typical usage case: browsing
Youtube) will regularly cause segmentation faults in Firefox for i386 users,
and will cause Flash content to "cut out" for amd64 users (nspluginwrapper
indirectly protects Firefox as it runs in an isolated process).

Although libflashsupport was disabled during the Hardy development cycle, it
was (perhaps unintentionally) left enabled for amd64 users, as the library
actually resides in the ia32-lib package (and Flash is a 32bit application).
The solution? The libflashsupport library needs to be removed for all
architectures and bug #198453 fixed. Flash 10 (currently at beta 2 in
Intrepid, with the release candidate available upstream) fully supports the
PulseAudio ALSA plugins, therefore making libflashsupport redundant.

Bug #190754 [3]. Several applications seem to have audio stuttering when
using PulseAudio output; for example, VLC and Skype (of which the latter
works only after fixing bug #198453). It may not be possible to fix this
problem for all configurations, but it is clear that tweaking the fragment
sizes in PulseAudio can make a positive difference for *many* users.

Bug 258581 [4]. A minor proposal to expand asoundconf's macros (which is
used to set the configuration proposed to fix bug #198453). The attached
patches (when revised to meet Daniel T Chen's requirements) will allow
slightly buggy applications to support PulseAudio when otherwise not
possible, e.g. Skype. This extended configuration is recommended by the
PulseAudio developers.

Additionally, two more bugs that are quite critical from an end-user's
viewpoint, but are not related to PulseAudio:

Bug #239182 [5]: There is a serious bug in Firefox 3.0.1 where "windowless
mode" flash content will cause segmentation faults in Firefox. This is a bug
in Firefox, fixed upstream and due for release in Firefox (actually
xulrunner) 3.0.2. If Firefox 3.0.2 (and nspluginwrapper 1.1.0 for amd64
users) does not make it into Intrepid, then we *must* configure Flash to
disable "windowless mode". If we don't, users will experience crashes on
very common sites, seemingly at random.

Bug #257403 [6]: This is the bug requesting an update to Flash 10 RC. I have
posted a debdiff which also disables "windowless mode", thus working around
bug #239182 without any significant loss of functionality in Flash.
I have set up my PPA for Hardy and Intrepid users to apply some of the
changes proposed in this guide. You can see my post [7] on the forums with
instructions for testing my packages and applying the proper configuration
as described on the bugs listed above, and you can also see some FAQ's for
PulseAudio on Intrepid [8] too.

I am more than willing to help contribute, and I understand there are
technical issues that will arise as a result of these proposed fixes, but at
the very least these issues need to be debated in a more focused way, rather
than spanning over multiple bugs.

Officially, we have eight days to apply the proper fixes, otherwise Ars
Technica will have to write yet another article as seen posted on May 8th
[9]. We can do better this time.
*NB: All of these configuration issues and bugs also apply to the Hardy
release. However, due to the strict SRU (StableReleaseUpdates) policy and
increased need for backporting / regression testing, perhaps we should leave
Hardy's issues for another day.

[1] https://bugs.launchpad.net/bugs/198453
[2] https://bugs.launchpad.net/bugs/192888<https://bugs.launchpad.net/bugs/192888>
[3] https://bugs.launchpad.net/bugs/190754<https://bugs.launchpad.net/bugs/190754>
[4] https://bugs.launchpad.net/bugs/258581<https://bugs.launchpad.net/bugs/239182>
[5] https://bugs.launchpad.net/bugs/239182<https://bugs.launchpad.net/bugs/239182>
[6] https://bugs.launchpad.net/bugs/257403<https://bugs.launchpad.net/bugs/239182>
[7] http://ubuntuforums.org/showpost.php?p=5587712&postcount=472
[8] http://ubuntuforums.org/showthread.php?t=866965
[9] http://arstechnica.com/reviews/os/hardy-heron-review.ars/4
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-devel-discuss/attachments/20080820/d7ec171e/attachment.html>