Sidebar: Development and Community

I'm typing this on my linux box for a change. It's appropriate because most of my Maemo experience to date has been through the browser and IRC client pointed to online maemo community chat channels and resources. This is a summary of my experience and impressions of the part of this particular PDA (yes i know that's a largely obsolete term...) that is unique so far: a vibrant online community of developers and users.

Take most of the complaints of this post with a grain of salt, as things are really in a state of flux due to consolidation of diverse Maemo web resources into a single destination (maemo.org). For example, the annoyance and confusion of having to register three times to access all the components of m.o will be resolved. Similarly, the unholy mess for developers and would-be "demoers" that is the scratchbox SDK is being slowly mitigated by other options (e.g. http://talk.maemo.org/showthread.php?t=38075).

Community
The N900 is an elite device, Nokia's "flagship" and the community surrounding it is fairly elitist as well. This makes for a lot of friction at the interface as ADD-affected kids and iJoes begin to trickle in to the fold. I experienced more than a little outright snobbery on IRC while beginning to grope my way up the developer learning curve. The good news is that amongst the socially inept and posers there are truly brilliant and helpful people who can save you hours of wandering in wiki hell. Wear your thickest skin and don't forget to google.

Which brings us to the wiki and the rest of m.o. The wiki is a mess, through no particular fault of anyone. If there was a single fault, it could and would have been fixed. As it is there's a combination of fast evolution (old, inaccurate information for previous Maemo versions is almost inextricably mingled with relevant Fremantle nuggets), fragmentation (beyond the fast pace of OS evolution, there are articles targeting all levels of expertise and intent), and above all poor visiblity and wretched search functionality that is still mostly broken from the aforementioned merging of Web resources.

So i've covered IRC and m.o and touched on the SDK. IRC is and always will be the niche of the nerds and we all want it to stay that way. The SDK experience and m.o are evolving quickly and, optimistically, upwardly. What's left is the actual experience of developing.

Development
I haven't done much software development and my CS classes were a long time ago, but a few things have already stood out to me as being fantastic about Maemo.

1. linux: like Android, Maemo is based on the linux kernel. The similarity ends there, however, and the differences are important. Android builds an application layer that looks and acts like a Java virtual machine on top of the kernel. This extra layer has some benefits but in my mind they are insignificant next to the restrictions. What makes linux so popular is not the kernel, but the kernel in combination with the "userspace tools"--all the ecosystem of tools, utilities and protocols that have been built up over decades of unix operating systems. Google shot themselves and everyone else in the foot by essentially throwing that out. With Maemo, any familiarity with linux or other unix systems will be a benefit. Web references for linux will probably be useful. Example code will probably be useful. Libraries and tools and apps will probably be useful. Seriously, there's very little reinventing the wheel here.
2. Debian: this is a more specific case of years of development and evolution in linux that benefit Maemo. The Debian package system and the tools that go with it are brilliant for simplifying the maintenance and distribution of complex software systems. The initial learning curve is more than compensated by the flexibility and robustness it brings to Maemo.
3. hardware: the N900 is loaded with good hardware, which simplifies things by making it work with a lot of apps already written for desktops. Sure, some UI and "optification" work may be needed, but having the grunt and the resolution to at least run these apps makes for a much broader world for would-be developers and maintainers.
4. No java: ok, this is extremely subjective, but i hate java and this is my own soapbox, so this is my little rant. It and the unrelated javascript have both been taken to places and repurposed for uses they never never should have been. Countless crappy software has been hacked together in both. Like any linux system, maemo system programming is done in C or C++. Python, as an elegant and easily-bound high level language, is also a viable option. Not having java supported in my mind avoids diluting the quality of available software by having cheap & dirty java versions of apps pop up that never work as well as the alternatives would--alternatives that either won't get made or won't get refined because the nasty java version is "good enough". I'm fully aware of the counter argument to this and my response is that i'd rather have fewer, better apps and that for cases where java is the only option, there are or will be solutions... :P

My own experience to date consists of the porting of uqm (The Ur-Quan Masters) to Fremantle, adding rudimentary accelerometer control through a joystick wrapper (which incidentally can be used for any other joystick-aware port--this is one of those cheerful little side effects of the three points above), writing an improved package description "nicify" function for the app manager (submitted as an attachment to a bug tracker enhancement request), and making an initial port of modern SDLMAME. The repository and hierarchy of extras, extras-testing, and extras-devel make it easy to provide a finished product to end-users while continuing work on a project and keeping it available for community testers and developers to play with. That makes it easy to work on a project when there is time, with the possibility of contributions from others who happen to have a speciality in some aspect of the development or have an itch to scratch. The setup and organization of maemo.org is really a big asset for Nokia, as long as both they and the community can keep making the planned improvements.

So, in summary, the N900 brings with it a huge community of developers and power users not seen for the Palm or Zaurus devices i've owned precisely because the platform development is actually driven by a community. While the members of the community and the tools have some thorns, they are overall becoming fewer. The nature of the OS and the sophistication of the device make developing for it a rewarding experience that can be as epic or minute as you wish.