I haven’t written much recently, because I’ve been busy. Work is prety hectic, and I’ve been trying to get some hacking done; I am painfully aware that I haven’t hacked on anything and released it for a little while now, and I don’t like that.
Unfortunately, the main thing I’ve been hacking on is an attempt to make my Zaurus be a media player (and not a PIM), and it hasn’t worked. After lots, and lots, and lots of trying, I have reluctantly come to the conclusion that it is not possible to write an application on OpenZaurus 3.5.3 which is the only app running on the Zaurus (or provides the only UI) and is in Python (because that’s what I can write code in). Note that I do not mean “an app running inside Opie or inside GPE“, I mean “it is the app running and that’s it; no desktop environment“. There seem to be three potential approaches:
- python-qt. To be the only app, you have to get launched instead of Opie. This requires python-quicklauncher to work, and it just doesn’t work at all.
- python-gtk. This involves starting an X server and then running your PyGtk app, so it’s basically the only thing running. This doesn’t work either; I can create a GtkWindow, but if I put any widget in that window, python-gtk segfaults on window.show().
- pygame. Doesn’t work at all; pygame has been patched to do something with the qt libraries which isn’t required any more, so it fails on a symbol import.
I’m really disappointed in this. I don’t want to rag on the OZ team too much, because they do a tough job (I’ve done a little work on the OZ website in order to make it a bit clearer, but I’m not part of the project) but, as far as I can tell, the packages aren’t tested much; pygame fails on “import pygame“, which means that it went into the feed broken and wasn’t tested. I’m not sure how to get around this without having unit tests for every package, mind, but when the alternative is “compile a new version myself“, involving setting up a cross-compiling toolchain, I’ve just knocked it on the head.
I have lots of other projects on the go, though. Hopefully some of them will happen, although more of them are involved with the Linux end of the world rather than the web end of the world, which is probably at odds with the relative proportions of people reading this, most of whom (I suspect) will be web people rather than Linux people.
However, whichever you are, you should be attending LugRadio Live 2005 because it’s going to be superb. There’s stuff for Linux people and web people and everyone in between. Go and look, you know it makes sense.