Mar 1, 2020

Open Source

I made a few small dejafu changes:

I tried doing a little refactoring to rule out some error conditions, but the resultant code was less clear. I suspect I’m at (or very close to) a local optimum with the current dejafu code. Any significant gains are likely to need a big rethink of the core approach.


No work this week. I’m back on Wednesday.


This has been a good week for lazing around and doing nothing of importance.


I finished reading The Unicorn Project. Like The Phoenix Project, which I read a few weeks ago, I wasn’t terribly impressed. It did fix some of the flaws of The Phoenix Project: TPP is about ops, and portrays them as pretty much infallible, merely misguided at times; whereas TUP is about devs and they are fallible. Well, except for the main character. To make everything right, you just need her to come in and, with a little functional programming and CI/CD, she’ll fix all your bugs while also making your code way shorter. The primary antagonist of the book is an exec who, it seems, hates success. Even when they manage the biggest promotion event of the company’s history, she’s all “we just don’t have the DNA to do IT, we need to outsource this before they run us into the ground.”

I’d say it was less bad than The Phoenix Project, but not by much.

I also finished reading The Return of the King. And I started reading The Vorrh.


My Apocalypse World game started this weekend. I’ve started a memo to write up the session notes.

I got and played Taur. It’s fun, but there are some aspects of the game which feel a bit unpolished:

  • Construction and research is done using gems which you get for completing missions. The distribution of these gems is incredibly uneven, I’ve got dozens of some and none of others. This means there’s often just nothing I can do to improve my base, despite the enemies getting tougher and tougher.

  • The game tells you that when some part of your base gets destroyed, you’re refunded the full resources. Sometimes that doesn’t happen, so things will be destroyed and (because the distribution of gems is really uneven) it often takes a while to build back up. So now you’re facing tougher enemies with a weaker base than one which lost before.

  • You can only construct or destroy parts of your base during missions, not between missions. This is a pain because, while you’re doing construction, time is still passing and the enemies are still attacking you! Albeit slowly.

I also got Middle-earth: Shadow of War. I’ve only played a few hours of this so far, but it’s been fun. It’s pretty similar to Shadow of Mordor, which I enjoyed at the time. For some reason Shelob, the terrifying spider-demon spawn of Ungoliant, appears as a human woman in this. I guess they felt a talking spider might be a bit much.


I watched all of Avatar: The Last Airbender, which was massively popular when it was on TV. I see why, it’s done really well. There’s good character development, very little feels like filler material (sadly, most of what did feel like filler was in the final season), and the world is interesting.

I then started on The Legend of Korra, which some say is even better. I’m only a couple of episodes into that, but it’s good so far.


I’ve been plagued, basically as long as I can remember, with great difficulty getting to sleep at night. Various theories have been put forth, none of which have really panned out:

  • Don’t have any caffeine after noon
  • Don’t use the computer for an hour before bed
  • Only use the bedroom for sleeping (eg, not for reading)

I’ve even tried taking disgusting-tasting (non-prescription) sleeping pills!

Since I started recording my sleep with an app about two years ago, I’ve racked up a 700-hour deficit, which means I’m under-sleeping by an average of one hour a night. Though in practice my sleep duration is usually between 8 and 5 hours a night, and the quarter-year average as of when I started my time off was about 6.7 hours a night.

While I’ve been off I decided to do an experiment: set no alarms, go to sleep when I’m tired, and see what the pattern looks like. My hypothesis was that I had somehow picked up a longer-than-24-hour cycle. Here’s the data:

Night From To Duration (hours)
Wednesday 0400 1200 8.00
Thursday 0300 1130 8.50
Friday 0300 1130 8.50
Saturday 0345 1215 8.50
Sunday 0315 1300 9.75
Monday 0345 1245 9.00
Tuesday 0445 1245 8.00
Wednesday 0445 1300 9.25
Thursday 0500 1315 8.25
Friday 0530 1230 7.00
Saturday 0530 1230 7.00

I’m not too sure what to make of this. You could make the case that the time I go to bed is gradually drifting later and later1, but the time I wake up isn’t experiencing a similar shift: they both vary independently.

One thing is clear though, I get much more sleep when I’m able to sleep when I naturally feel like it.

  1. Sadly I don’t have a month or two to test this hypothesis, and as I’m back at work on Wednesday I need to try to abruptly shift back to waking up at 7AM.↩︎