To my mind I would rather see Nokia spend effort on infrastructure, development platforms/tools and documentation rather than the "killer"apps. Since they obviously have thought of this as a platform I don't see why they should have to write all the apps. I think its our job here to come up with what we want and somehow have some means of communicating that to the development community at large, rather than putting it all on Nokia's shoulders.

I think Nokia has done a very good job with the development environment, it seems light years ahead of what has to be dealt with on other handheld "platforms" . We need to make use of that and figure out a good interaction model with the non-Nokia developers to get the applications we want. I think that is the critial piece to this puzzle. This frees up Nokia to work on what only they can do (like getting OpenGL drivers ).

One thing that I think would help a lot is an easy to use simple script packager so you don't have to be a full fledged maemo developer to write a script for the IT. This would take a script (say shell or python) and package it up so it can be installed with the application manager and run from the application launcher (setuid capability would also be nice, but that gets a little tricky).

A case in point: accessing windows shares. This is doable on the IT, the CIFS kernel module exists and documentation for use also exists. BUT its a pain in the neck to use, especially for non-linux hackers. First you have to get xterm, then figure which of 16 ways to get root access you are going to use, figure out how to use the command line to get around (ls, cd etc), understand the difference between user space file structure and linux file structure, then properly run all the commands. Even if you write a script for this you still have to start xterm and become root to run it. We need some way that this simple thing can get scripted up and used just like another tool. You shouldn't have to learn to be a full fledged maemo developer to do this.

This one simple thing I think would greatly accelerate the spread of usefull little applets amongst the user community. Take for example the schedular mentioned previously. A simple schedular could be written in 20 lines of python, it wouldn't be terribly powerful, it would provide some utility and would be easy to write. Someone else could come up with ideas for enhancement, tweak it a bit and an even better version would exist. If a lot of people really like it and use it a development group could take over. I've seen this happen in many other environments. I don't see that happening here, primarily I think because its so difficult to use such scripts.

Well thats enough for now, I have more to say, but I'll save that for later.

John S.