Mobile linux and the desktop

Edward pointed my to this blog today which brought up a point I myself have been making in regards to Android. I spoke to several people at the CE Linux meeting a couple of weeks ago about this for one. To quote from the blog:

Android is an island of its own, and useful code sharing is largely limited to the kernel.

At Collabora Multimedia we are currently working with both Maemo and Android systems and while I can see the appeal of Android from a phone makers perspective I can’t help but be a little saddened by how worthless it is to the general linux eco-system. One of the things I always loved about Nokia’s Maemo effort is that since its using so many of the standard components that we use on the Linux Desktop, it means that when a feature is added or a bug is fixed in Maemo, it directly helps also the linux desktop. Nokia and Maemo has had a strong and direct impact on a lot of open source projects, ranging from GStreamer, D-bus, GTK+, Telepathy, Matchbox, X Window System and more. And Nokia’s work on Qt going forward will of course have a direct impact on the quality of KDE.

Android on the other side has a much more marginal impact. I know they have contributed some patches to Webkit, but apart from that they offer little value to the rest of the linux eco-system. Been even told by some kernel developers that an Android kernel driver is about as immediately useful for the mainstream kernel as a FreeBSD or OpenSolaris driver. Meaning that porting is needed.

So for me personally I can’t help but feel a lot more positive about Maemo (or Moblin for that matter as they too share the same kind of philosophy as Maemo) and getting a N900 is definitely on my TODO list. That said Android is a work in progress and hopefully we can get them to abandon their essentially proprietary stack going forward and instead incorporate more and more shared libraries with the server and desktop. Maemo has proved that for a smartphone these libraries works just as well as Googles homebrew. Some of the efforts we are involved with are pushing in that direction and hopefully Google will realize that the secret to the success of open source is synergy.

15 thoughts on “Mobile linux and the desktop”

  1. Same here. Android only provides a benefit to the phone makers and those selling the phones, because they can lock it down. Maemo in contrast benefits the free desktop, and the power-users who want to have a high level of control over their devices. Considering Intel’s plans for 2010, it seems that they are planning to provide chips suitable for smartphones and intent to bring Moblin to the smart phone.

    I also plan to buy a N900, not only because I need a music player and a mobile phone (then I could have bought a cheaper one), but also because I can do so many things with it like developing software easily (OK, scratchbox is a bit complicated to install, especially on an amd64 system, but it’s not very hard).

    So, basically the difference between Maemo devices and Android devices is that the first one runs GNU/Linux and the second one not. Moblin devices also run GNU/Linux. Thus we can say that all devices running GNU/Linux benefit the free desktop. Thus who just run Linux with a different userspace are almost completely useless.

  2. Palm WebOS should hopefully become a useful contributor to the Linux mobile space as their stack is mostly a gnome mobile stack as well using gstreamer and the like just like Maemo and Moblin.

  3. I agree that the current (and probably future) situation regarding Android is not ideal.

    Then again, how many devices use Android vs Maemo? Right. Maemo is Nokia only and there is only a single device running Maemo5. Android on the other hand is used by a lot of companies and ships on very different hardware.

    As much as I would love Clutter and Telepathy based handsets being shipped by a lot of manufacturers, it won’t happen.

  4. @Michael: Maemo might be Nokia only, but its contribution to the community is still crushing Android, also Nokia only is still a huge chunk of the market. And you have efforts like Moblin and Palm WebOS which are very similar to Maemo. And while there are talks about a lot of stuff, so far I haven’t really seen any Android devices apart from the two from HTC.

    But in some sense the important part here is not size of marketshare. The value of Maemo and Moblin are much greater to the open source community no matter if their market share is 10% or 80% as things currently stands. While Android is currently of very little value to the community regardless of Market share.

  5. You mean abandon a stack that they are busy selling applications for right now and apparently does what they want? Yeah…

  6. That’s quite unfortunate to see that people are only talking of big players.

    There is other stacks such as pokylinux ( from openhand, bought by intel for moblin http://o-hand.com ), there is lots of stacks directly based on linux for openmoko phone ( like hackable:1, based on gnome-mobile , made by bearstech ), there is indeed webos.

    And I think most of them offer a standard linux api when they exist. For example, tangogps, a gps application made for openmoko phones run fine on a laptop with a standard gps. Midori/arora can be used on a mobile device. The whole telepathy stack and gstreamer stack is used on most plateform.

    And that’s right the synergy is the key.

  7. @Rob J. Caskey: an advantage of Android is that Google so far has made no promises on the stack beneath their JAVA API. So they can in fact move in a more standard direction with Android without abandoning their current apps. That said the idea that Android apps currently truly works nicely across all devices and resolutions is a illussion.

  8. As Android needs a working base system, with a kernel, a GUI, a multimedia system, etc. I’m wondering if simply porting their java VM (Dalvik) to e.g. Maemo would work ? That would be a great platform.

  9. The Google Chromium team is contributing a lot, the Android team fails to participate in upstream development, send patches, open bug reports, leave a lone producing a clean patch from android to WebKit.org..

    Their git branches is a dump from perforce to git and figuring what they actually changed and what they cherry-picked from webkit is quite hard.

  10. @Alex: Android like any opensource project does contain some useful code. But the opencore-amr libs wasn’t as much given to the community as it was something the community was able to salvage out of Android. I mean the AMR files where not made public in a way that was useful to anyone outside Android, so it was only due to the effort of the people behind the opencore-amr project to take that code and make it something that could be shared by a lot of projects.

  11. In the Android stack, Linux is a mere hardware abstraction layer for the virtual machine running on top of it and supporting cloudy applications that keep the user close to Google.

    I have never been fond of Nokia’s software (great hardware though) and I’m suspicious of something supported by a single vendor, but Google’s choices with Android have made me see Maemo in a much more favourable light.

Comments are closed.