By Jake Edge
November 23, 2011
There have been rumblings for a few weeks now about the fate of the Banshee
music player in the Ubuntu default install. The distribution switched from
GNOME's default Rhythmbox music player to
Banshee for the 11.04 release,
but started discussing switching back for 12.04 LTS at the recent Ubuntu
Developer
Summit—to the surprise of the Banshee team. While the reasons for
the switch are largely technical, though there is some dispute about their
validity, there is a fair amount of unhappiness about the
decision, not only for the switch itself but for the way the decision
was made. In any case, the distribution will be switching back to Rhythmbox
for 12.04.
One of the biggest complaints is that the Banshee team had no warning that
Ubuntu was even considering the switch. As Jo Shields put it in a blog post: "If
Banshee was being considered for replacement due to unresolved technical
issues, then perhaps it would have been polite to, I don't know, inform
upstream that it was on the cards?" He also notes that this is not
the first time a project has been blindsided by a UDS discussion to remove
a package from the default install. The PiTiVi video editor project had a similar
experience back in May with regards to removing that package for 11.10.
After the UDS meeting, Canonical's Ubuntu desktop manager Jason Warner posted a note to the ubuntu-desktop mailing
list looking for "further feedback". He outlined some of the
problems with Banshee that were brought up at UDS, including its stability,
start-up time, resource usage, and the project's responsiveness to
Ubuntu-specific concerns.
Ubuntu Banshee maintainer Chow Loong Jin posted a
response that addressed each of those concerns, but seemed most puzzled
by the lack of responsiveness complaint: "If there were patches that
Canonical/Ubuntu wanted in, why haven't I heard about them?".
Martin Pitt concurred that Chow and Banshee
have been responsive, but he added another handful of concerns about the
program, including GTK3 support, the amount of space required to ship
Banshee (30M, which includes Mono), ARM support, and a "fuzzy"
future for Mono.
Once again, Chow stepped in to address some
of those complaints. In particular, there seems to be a disconnect
regarding ARM support. Chow points to one bug that has been filed, which is
being worked on, and is unaware of other problems. But, based on information from Tobin Davis and Oliver Grawert of the Ubuntu ARM team, there
have been numerous problems supporting Mono and Banshee on ARM. Those
problems have evidently taken a large portion of the ARM team's time to try
to resolve. That, coupled with a perception that Mono is becoming less
popular as a platform, has even led to consideration of dropping Mono from
the packages that will be officially supported for the LTS as Sebastien
Bacher mentions:
We also
didn't take any decision on the topic but if mono is having low traction in
the linux desktop world nowadays (it was pointed at UDS that we don't see a
lot of new projects using mono since f-spot, banshee and tomboy, most new
GNOME projects seems to go for vala rather for example) the cost of
supporting it officially for 5 years might be higher than the benefits,
it's not a made decision of any sort but a topic to discuss and take in
consideration there.
The technical complaints seem reasonable, though it appears that some of the
stability, resource usage, and start-up time issues are debatable, with
some having lots of those kinds of problems and others not seeing them at all.
But the bigger question is why the project wasn't really notified that
there were big enough problems to consider removing Banshee as a default
application
until after it had been proposed, and mostly agreed upon, at UDS.
One of the Ubuntu Mono packagers, Iain Lane, was actually present at the
UDS, but was evidently never notified that Banshee's removal was under
discussion. As Lane put it :
None of this should have come out of a UDS session. If there are
such serious issues then they should have been raised months ago.
Lane is reacting to Warner's pronouncement, two weeks after bringing the
subject up, that "it is clear the right decision for 12.04 is to make Rhythmbox
the default music player". That message, in effect, brings the discussion to
a close and makes the switch back to Rhythmbox. Lane and others felt that
it was far from
"clear" that the switch should be made. Warner clarified that he meant "clear to
me", and went on to specify the reasons behind his decision. Most
of it comes down to stability:
Most people using the traditional "Linux Desktop"
are vocal, though forgiving, when it comes to applications. I firmly
believe that mass market users will neither be vocal nor forgiving and
they will not tolerate systems they can't trust or feel are unstable. They
simply won't use our products and apps.
IMO, the default apps need to be the best foot forward, the showcase for
Ubuntu. If a users first experience with our product is poor, we have a
problem from which it will be very hard to recover. Rhythmbox, while not
pretty or as featureful as Banshee, does the core function more reliably
and more stably. And it isn't about number of features or [breadth] of
feature scope, but rather how well the application does what it says it
does.
For his part, though, Lane also notes the PiTiVi situation as an earlier
example, and details the problems he sees in
the process:
Most importantly, we need to be having more constant and constructive
dialogs with upstreams throughout the whole cycle and not just at
application selection time (actually in this case there has been very
poor [communication] with upstream even since the discussion). Using the
promise of being on or threat of being removed from the install to
cudgel projects into the direction you want is not very satisfactory.
Chopping-and-changing doesn't do people any favours either. There should
be some commitment from Ubuntu; being the default app brings upstreams a
lot of users (which is great), but also increases the support and bug
workload (which is not so great). If Ubuntu could be relied on more
then upstreams would not be so burdened.
Bacher acknowledges some of Lane's
complaints, saying that the "topic has been handled in a pretty
suboptimal way". He also agreed that the PiTiVi situation was
similar and that the distribution could have done a better job keeping that
project in the loop. But he is leery of Lane's "cudgel" characterization:
While it would make sense to tell upstream what we like or dislike
about their software, I'm not sure we should try to "use" our position to
demand things to be done or fixed for us. I would feel uncomfortable
telling any upstream "you should fix those bugs or your software will be
out of the CD", while in practice it's true that some issues lead us to
decide what to ship or not.
There's also the question of the importance of being in the default
install. Even if Mono were to be removed from the packages supported for
the five-year LTS period, both it and Banshee (Tomboy, F-Spot, ...) will
still be available, but may be less prominent. As Lane pointed out, there
are both good and bad points to being in the default install, but it's clear that projects which get removed
feel it to be something of a slap in the face. But, as Jono Bacon points
out, the way that Ubuntu users discover new software is changing:
While a few years ago apt-get and Synaptic provided a means to install
software for us, they lacked the discoverability and ease of use required
for many novice users. In addition to this, those tools did not provide a
means for a novice to identify what were the best choices for them in the
sea of software available for Ubuntu. As such, for software not included in
the default install, end users were somewhat in the dark about the best
software they could install to meet their needs.
The Ubuntu Software Center changed all of that. Now we have a simple, easy to use facility for browsing software, seeing ratings and reviews to get a feeling for what is best of breed, and installing it with just a click.
The Ubuntu Software Center
provides an AppStore-like interface for users to discover new software, so
they are less likely to be bound only by what's available in the default
install, according to Bacon. However, he does recognize the importance
of the default applications and why projects are concerned when they get
dropped:
I am not surprised that some consternation occurs when applications or
components are proposed to be added or removed from a default Ubuntu
installation. Being on the disc provides a sense of validation and
acceptance to our upstreams, and provides an incredible amount of
visibility to these applications.
In the end analysis, it would seem that Ubuntu could do a much better job
working with the upstream projects that it is shipping in its default
installation. It is mostly just a matter of communication, and this incident (and that PiTiVi situation before it) have raised the
visibility of this type of problem. Rather than waiting until (nearly) the
last minute, involving the upstream projects as well as the Debian/Ubuntu
package maintainers early in the decision-making process is likely to cause
a lot less stress and unhappiness. Projects that know about problems that
may lead to
their eventual removal from the default package set are pretty likely to
address them—giving them enough time to do so before suddenly tossing
them out can only help with that.
(
Log in to post comments)