Pulseaudio dependency, if Debian can do it ...

[Daniel Chen]

Hi,

On Sat, Jun 25, 2011 at 4:21 AM, rosea.grammostola
<rosea.grammostola at gmail.com> wrote:
> 1) Fedora, OpenSuse and Debian etc. aren't building their distro especially
> for this group.

Not being involved with Fedora or openSUSE closely, I can't comment on
them, but Debian doesn't ship a desktop as tightly integrated with
PulseAudio because the objective of offering a streamlined desktop
environment based on GNOME isn't the top priority.

> 2) Why make it more troublesome for a group of users when it is not needed
> and other distros are showing it can be done without hurting the usability
> of the Desktop?

There has never been an intentional effort in Ubuntu to sabotage the
removal of pulse, rather there were no resources to better integrate
its alternatives (be they a complete removal of pulse or a drop-in
replacement, which to this day, still doesn't exist).

> 3) Shouldn't components in the Desktop be able to be removed as much as
> possible? Just as a matter of Unix / Linux principle or doing things right?

As a matter of principle, this approach is ideal. Pragmatically it's
not straightforward at all. Consider the case of removing ALSA and
replacing it with OSSv4: it's not for the faint-of-heart and requires
a tremendous amount of corner-case awareness not to let certain
applications fall through the cracks. Or if you'd rather look higher
in the software stack, consider the case of replacing Evolution with
another calendar+email app that retains indicator functionality.
"Doing things right" /could/ imply providing full functionality or
being a drop-in replacement, but those approaches might not mesh with
the rest of the environment, and then you're faced with changing the
entire environment.

> 4) Maybe it's not your focus group of users and you don't care, but you
> probably don't have an idea about how many people in the Ubuntu community
> are working very hard to make Ubuntu good for music (home)recording, via
> bugreports, package building, PPAs, Ubuntu Studio, documentation etc.

No, you're right in that I don't have an idea about how many people in
the community are working on improving the distribution, but I don't
believe that is the objective here, which is to contribute those fixes
back to the appropriate level. Things that improve pulse as a whole go
to upstream pulse; things that improve the base packaging go to
Debian; things that improve the integration into Ubuntu go to Ubuntu.

The idea is not to strike Ubuntu as a choice but to realize that this
distribution may not be the ideal choice. As you allude to later, it's
certainly possible to make Ubuntu do things, but the distribution
itself fits on one 700 MB CD image, and in those constraints we must
consider the likely computing needs of an audience. That said, it's
possible to create metapackages that contain information about
conflicting with other packages or providing certain functionality
(I'm touching a bit on the Debian packaging terminology), but to
create a streamlined solution like (the original) 64Studio really
requires things like a realtime kernel, the assortment of JACK/2-based
applications, and so on. That's actually the intent of the Ubuntu
Studio derivative.

> Fedora uses GNOME, Arch Linux uses GNOME, OpenSuse uses GNOME, Debian uses
> GNOME etc, but it is still good possible to remove Pulseaudio. Apparently
> those developer teams are seeing an advantage of the ability to remove
> Pulseaudio. And I think, even if there was not a obvious reason why people
> should want that, that is a general good and clean way to handle things in
> the world of Linux Desktop.

I don't think those developers necessarily see an overt advantage of
removing pulse. As far as I know, none of them ship indicators in
their default environments, and indicator-sound is a significant part
of the default Ubuntu experience. Arguably indicator-sound could be
extended to work directly with the ALSA hw ctl layer, and if you know
people who are willing to prioritize that use case, I'm sure Connor C
would be happy to discuss the merits and drawbacks of said approach
and move forward.

> The fact is that a group of users wants to be able to remove Pulseaudio. The
> question is why this is possible on other GNOME distros but not on Ubuntu?
> Is there a way to make this possible on Ubuntu also? Are you willing to make
> this possible?

I presume you're asking whether it's possible to integrate methods to
make it feasible "with one click or something close to it," and if so,
yes, there are people willing to work on, but we need those people to
step up and act more visibly with the Ubuntu development team. The
ubuntu-audio-dev team on Launchpad is a good place to begin detailed
discussions.

Cheers,
-Dan