7 comments

[ 4.4 ms ] story [ 34.8 ms ] thread
Wolfram's graphs were prettier :)
True. He does have about a lot more data over a 40x longer time interval, so the noise smoothes itself out more.

Unless you also mean I should have formatted the matplotlib graph better? Mathematica actually has quite nice plotting in general. I also wanted to use d3.js but that sort of defeats the purpose of personal data.

The parse.awk script fails on Linux (no -j option to date). I didn't manage to decode what the date command does in the awk script, so I just used the python script instead.

It might be interesting to allow parsing for specific events rather than just when the kernel is logging vs. when it's not.

OS X uses FreeBSD date which has totally different cmdline options. I did actually write an awk script that works on Centos 6, I'll upload that shortly. (Edit: I think instead of -jf, you can use --date= on linux.)

The event suggestion is good. Really I just threw the script together last month and occasionally fixed it up when I showed it to friends, so maybe I'll take some time to add that and fix some other things about this (like a prettier graph ^) when I finish my last stretch of schoolwork in the next few weeks. But in the mean time feel free to add that if you want :)

I'll take a look this evening at trying to get it to parse specific events. Like you, it's busy times at the moment.