Recent blog entries for micahjd

This stats association graph for CIA is starting to show some interesting results. I've had the 'real' CIA server collecting data for a few days now, and I've just started playing around with it using graphviz, python, and sodipodi.

Anyway, you can expect soon to have a system for automatically generating graphs like these in the CIA web interface :)

29 Jan 2004 (updated 29 Jan 2004 at 20:12 UTC) »

Well, I think I'm finally over the insomnia I've been having the last few weeks- not being able to sleep, waking up at 2:00 AM, knowing that my brain's energy was completely depleted but not being the least bit tired. Glad that's over with, I hope.


I'm still planning in my mind the last few pieces necessary to do properly generic association between stats targets- finding all the authors working on a project, finding all the projects a person works on, hyperlinking messages to other stats targets, all without having a hardcoded definition of what an 'author' or a 'project' means.

Luckily, there's no shortage of incremental improvements to be made- lots of small enhancements like better caching and automatic thumbnail generation.

There is a publish/subscribe system implemented using the "cloud" tag in the RSS 2.0 specification. This lets users subscribe to get instant notifications when new data arrives on an RSS feed. The server works, and I have a proof-of concept client that works, but aside from Radio Userland I haven't found any RSS aggregators that support it. Hopefully I'll try writing a patch for Straw soon.

One of the niftier recent developments is that every commit message now has a full web page associated with it, containing the unabridged commit log, a tree of modified files, and various other tidbits. Gnome commits even get a hyperlink to Bonsai.


Not a lot going on... this AI class is starting to get interesting. It's the second AI class offered by the university- the first one was a pretty good overview of the field, but this one is focusing pretty much exclusively on Support Vector Machines- a rather cool machine learning algorithm based on finding a hyperplane that separates your data when expressed as vectors in a very high dimensional space. SVMs are a pretty new field, but already seeing lots of use in bioinformatics and handwriting recognition.

Aside from my AI class, it's all pretty mediocre. Math that I've known since high school, "writing for engineers", an intro psychology class that's covering less than my high school psych class. Ah well, it's all for those humanities credits you seem to need to get a CS degree.


On Monday I finally sent out the Critical Decoder board for manufacture. With a little more software and a lot of luck on the CSGC's part, this thing should orbit the earth within a year. Kinda neat.

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!