The first two days at linux.conf.au are dedicated to "miniconfs," which
cover specific areas of interest. The 2009 event in Hobart, Tasmania
included a miniconf for mobile Linux; your editor attended a few talks
there. As might be expected, there is a lot going on with mobile Linux,
and a lot of interest.
Nancy Mauro-Flude is a performance artist who has used mobile Linux as part
of a device intended as an artistic and political statement. The Baglady device
is a purse with a numeric keypad on the outside. Inside, it contains a
Linux-based system with wireless networking. A camera and microphone have
been discreetly placed on the strap.
When enabled, this device captures pictures and audio from its owner's
travels, then immediately uploads them to a remote server. It allows its
owner to capture the events around her, perhaps in situations where
recording devices are not appreciated or allowed. The immediate-upload
feature ensures that the data gets out, even if the device is discovered -
at least, in places where an open access point is available.
The subversive possibilities of such a device are clear; so are the
potential privacy problems. Nancy was clearly aware of those issues, but,
arguably, has not worked through them completely. Others will certainly
follow this particular artist's lead; expect to see more mobile devices
which record their immediate environments and put the results on a server
for all to see. It is going to be interesting.
Canonical's David Mandala gave a well-attended talk on Ubuntu's efforts in
the mobile arena. Like other such projects, the Ubuntu Mobile
effort faces challenges beyond simply making the distribution run on mobile
systems. Mobile systems truly are different, and, as a result, a user's
expectations of the operating system are quite different. Small screens
are a problem; not all applications have been written to function well when
the amount of screen space is limited. Touchscreens complicate things
further; David issued a challenge to developers to find ways to allow more
space in menus so that fat-fingered users can use them on touchscreen-based
The Ubuntu Mobile effort is actually two related projects: Ubuntu MID (for
small, tablet-like devices) and the newer Ubuntu Netbook, aimed at larger
devices. The Ubuntu MID work is currently based on GNOME Mobile, though
David suggested that things could change at that level. In particular, he
said, the Qt license change has stirred things up a bit. There is a
selection of applications which are optimized for small screens. The
distribution as a whole is intended for original equipment manufacturers;
it is not expected that users of MID devices will be installing their own
MID systems typically use a touchscreen as their primary input device.
Netbooks, instead, combine a larger screen with a real keyboard; that leads
to different requirements. The Ubuntu Netbook distribution uses the full
GNOME desktop - for those applications which behave well on an
800x600 display, at least. This distribution should be available in stable
the end of the Jaunty development cycle.
David seemed to be having the most fun, though, with the new Ubuntu ARM
port. One does not normally think of the ARM processor when one ponders
netbook devices, but it seems that ARM is making a real effort to enable
products in that area. As part of that work, ARM is working with Ubuntu to
have a proper distribution ready. This effort seems to have gone pretty
well; at this point, the full Ubuntu distribution is available for ARM
systems. The biggest difficulty, it seems, is that ARM-based systems lack
proper video acceleration. Canonical is working around this issue, though,
and plans to support this port along with the others.
It seems that Canonical sees a bright future for the ARM port. While there
are a number of systems available for x86-based devices, there is no real
competition to Linux on the ARM processor. Windows does not run there.
Symbian does, but it is not a true desktop-based system. So, any ARM-based
netbook devices which appear on the market are sure to be running Linux.
Canonical is doing its best to ensure that they run Ubuntu in particular.
An alternative for small systems is Poky
Linux, a system put together by Opened Hand prior to its recent
acquisition by Intel. Poky Linux is, in fact, two different things: it is
a system for building Linux-based platforms, and it is also the
distribution which is that system's output. Rob Bradford, in his
presentation, acknowledged that this naming practice may lead to some
confusion. Still, while Poky may suffer from some ambiguity, its
developers seem to make up for that with enthusiasm.
Poky Linux started as a fork of the Open Embedded platform. The
developers tossed in a bunch of tools which are useful on small devices:
the Clutter desktop work, GeoClue, the "Sato" user interface, the Pimlico personal information
management system, GStreamer, WebKit, etc. The result is a fully-featured
distribution which is well tuned to the small device environment.
Perhaps the highest-profile use of Poky Linux is in the Vernier Labquest device.
Rob discussed at length the build system that was created to allow the
creation of Poky Linux distributions. There are a lot of tools there which
make the task relatively easy, and which, as Rob pointed out, are well
suited to people who do not like to type very much. More information on
how that works can be found on the Poky
What the audience really wanted to know, though, was Intel's intentions for
Poky Linux, which it acquired with Opened Hand. Though Rob didn't say so
directly, the real answer appears to be that Intel doesn't have much
interest in Poky Linux and is not putting resources into its further
development. So, says Rob, while the infrastructure is still in place,
Poky Linux has become a community project. The future of this project, it
seems, is in the hands of those who use it and wish to see it continue.
GeunSik Lim gave a talk outlining the internals of the Android system.
Much of that talk is not amenable to summarizing here, though there were
useful details which will help as your editor digs more deeply into that
system. One thing that jumped out, though, was this: Google decided to
create its own C library for this platform. The size of glibc was
part of the motivation for this work, but the real reason, it seems, is
that Google doesn't want to have GPL-licensed code running in user space.
They worried, perhaps, that glibc could go to GPLv3 in the future; that, of
course, would make it impossible to use in a locked-down device. So they
started with a BSD-licensed libc which was then tweaked extensively for
their needs. The resulting library (called "Bionic") has some big gaps (no
support for C++ exceptions, for example), but it evidently suits the
Android platform well.
In summary: mobile Linux is clearly one of the hot topics for this year.
There are a lot of people and projects working in this area, doing no end
of interesting things. It is going to be fun to see what our community
comes up with.
to post comments)