So yesterday was a contrast as I spent the majority of the afternoon re-writing my quake 4 log analysis perl script & manually checking the data. Don't think I'd mentioned it here before? I discovered that the xbattle mod logs out the sort of data I wanted to analyse - quite like the quake 3 logs, in fact, though there're some differences. kleeks (from the jolt servers where we normally play of an evening) kindly gave me a log from their xbattle server to give me something to work with and I'd got some basic stats done (here, but I don't think there's anyone that people who read this know) but next on the agenda was manually checking a couple of games from that log just to be sure I'd got it right. And instantly I discovered a mistaken assumption I'd made about how things were logged. Which meant gutting the script & re-writing a fair chunk of it. But it's now simpler and I think I've ended up with a better way of doing things, so it's all good in the end. I think it'll've changed the figures slightly though.
I then spent a large chunk of the afternoon going through 17000 lines of log (just 2 games worth) counting kills & checking name changes/disconns/reconns were working properly. Of course the numbers didn't match - I'm missing about 3 frags out of 300 or so. I think that's my error rather than script error though - there doesn't seem to be anything systematic about it. Double checking my counting should thankfully be quicker than first time round, I only need to look for kill/suicide/worldkill lines as the other stuff checked out fine.
Then I need to make it deal with multiple log files, then I think I'm done with dull bits for a little while. And can add some more stats :) One thing that does need to be done longer term is read up on cgi scripts & make it a little more elegant - at the moment it writes out umpteen html files, it works but it's somewhat cluttered. But that's for a latter date, I think.