3 min read

Weeknote 2021.08: Sweat the small stuff

  • Not much to report again this week. Still waiting on keyboard PCB’s so no change on that front.

  • Did knock up a quick 3d print case design loosely based on the old IBM F62 Kishsaver design. haven’t finished printing and assembling it yet so no pics for now; I don’t want to ruin the surprise.

  • After a bit of an abortive start last time. I started to run again. I’ve scaled back a bit and been a bit more thoughtful about the process. I’ve calendared in 3 runs a week, on Monday, Wednesday, and Friday, with a tentative cycle ride on Sunday. I’m also running a much shorter distance to start with and am alternating periods of running and walking so that I can build up my ability slowly (in the same vein as couch to 5k). I’m already seeing some improvements after just a week which I’m super happy about.

  • I have a problem with detail. This reared it’s head again this week when I spent almost 3 days debugging a problem that ended up being caused by my use of the wrong bitmap during one of the mark functions (for the curious: in Rubys GC, HEAP_MARK_BITS are not the same as HEAP_MARKING_BITS).

    Predictably as soon as I asked for help and paired on this problem we spotted it within 20 minutes and I felt like an idiot.

    After feeling (rightly, imo) angry about the confusing and slightly disingenuous naming patterns in the MRI source code, I pushed that thought away and realised that this was just another instance of a pattern of behaviour that has haunted me for years.

    I can be detail blind, and I don’t know what to do to fix it.

    Within a work context this often manifests in one of two main ways:

    • I finish a feature, push a PR and then spend days fixing feedback from my colleagues in the form of small changes that often seem like nitpicks. A misleading variable name here, a typo there, a slightly obtuse algorithm in that other place. Or
    • Upon discovering a bug, I go down a rabbit hole for a long time, exploring the problem, checking the algorithms and high level approach, making sure that the test cases are thorough only to discover that the bug was caused either by a typo, or by a trivial mistake that in hindsight should have been obvious.

    I know I have a tendancy to zoom out and focus on slightly higher level concerns. When I’m building a feature I focus on whether the approach makes sense, whether I’m using appropriate patterns and data structures, or whether the object graph makes sense, so much so that I lose sight of the smaller detail. This seems to get more acute the higher mental pressure the work has, so whetherI am fixing a problem in my code? someone elses code? refactoring a feature or building a new feature (probably in that order, high to low).

    Anyway. I haven’t worked out what to do about this yet and it sucks because it’s embarrassing and it’s holding me back.

    But the first step to fixing a problem is admitting you have one, right?

  • I updated my User Manual.

    I’ve been keeping this up to date as I move between jobs and as I learn more about myself since I was working at FutureLearn. I don’t think I’ve ever shared it publicly before, but seeing as this week is a slow news week I may as well!