2010

Sat, Jun 26

<14:22 EEST> In recent news, Google makes it clear it is no better than Apple, when it comes to the question who owns your mobile phone. This ties in closely with my January musings, as well as the more general rant on locking away perfectly good hardware. It seems the only sensible possibilities for a geek's phone right now are either the Nokia N900, or a simpler, more fixed appliance.

While the N900 and Maemo/Meego are not perfect, there are two important things going for them. Unlike Android, where applications are managed Java blobs, Maemo is pretty much like any Linux distro. If the multiple hacker-oriented repositories are not enough, you can always install development tools and compile stuff yourself, using whatever language you like. I am not familiar with Meego, but at least you can keep using Maemo — or some other OS, as there are multiple alternative distros available.

So even if Nokia starts closing things up in the future, which is not all that impossible, considering the lack of truly open competition, the N900 is still open. This is in stark contrast to Android and iOS (of iPhone/iPad) where the parent company can remotely take control of the hardware you think you own.

Given such a bleak state of the smartphone scene, and the steep price of the N900, my appreciation of the 'old' Symbian generation is somewhat increased. I consider these the 'simple and robust appliances' mentioned in January, even if they still have relatively complex and flexible software. For one thing, Symbian seems to lack the big-brother mentality. Besides, an N900 might be a little too big for carrying with you all the time, yet a little too small for a decent computer.

There is also a stupid trademark issue behind my loathing of Android. Google asked the Finnish company Zendroid to change its name because of a supposed name confusion. One of these companies deals with robotics that strives to emulate human tactile and motor behaviour, the other does not.

Thu, Jun 3

<23:57 EEST> This is a kind of mental note that the appletouch driver is broken in Linux 2.6.34. It worked in 2.6.33.x, and the old appletouch.c file can be used as such in the new kernel to get a working driver. At the moment I lack the energy to start reporting on this, as I have other things in mind, and pointing out bugs/regressions has turned out surprisingly difficult in the past. However, somebody may pick this up and share the pain, so hopefully we can get somewhere.

The root of the bug seems to be that appletouch is used for several slightly different touchpads. The change in 2.6.34 fixed a (literally) edge case in the latest Powerbooks, but the touchpad in my slightly older Powerbook goes completely erratic. About one quarter of the touchpad works normally, while some areas work in inverted directions. Looks like a genuine case of broken signed/unsigned type casting, if I've even seen one ;) There is only one other complaint I have encountered so far, but distro kernel users will probably have fun with this later.