Weeknotes 2022-24

This week’s main news is that Shadow cat has gone missing again. She didn’t turn up for any food on Wednesday, and I haven’t seen her since. She did disappear for a week a year ago, and turned out to have been hiding in my garage the whole time, but I’ve checked several times and she’s definitely not there this time. Still holding out hope that she might reappear again, but obvious the longer she’s missing the less likely it gets.

I put together a cryptic crossword for the first time in a while, and people seemed to largely enjoy it (the kind of people that have any interest in such things anyway), so I should probably do this more often. This one had a theme, which I usually seem to end up doing – you’d think it would be easier without, but it feels a bit weird for some reason.

I spent Monday afternoon grappling with a failing build caused by race conditions as some tests tore down database sessions after doing things which published messages to a queue, which caused things listening to those messages to terminate unexpectedly. There’s probably a lesson there about decoupling unit tests from the database, but in the meantime I tried all sorts of tweaks without any success. Then just after I’d given up for the day José Valim retweeted a tweet from German Velasco linking to his blog post about how Wojtek Mach had fixed a similar issue in LiveView. It wasn’t exactly the same situation as I had, but it was enough of a nudge in the right direction for me to finally fix the issue the next morning.

I did a quick TDD demo for some colleagues. It’s always hard to pitch this kind of thing so it’s neither too simplistic nor so complex as to be impenetrable or boring – this time I prepared a simple to-do web app, then test-drove a made-up feature involving a button which popped up a random item out of all those with the highest priority which weren’t yet marked as done. This meant I could throw in a bit of UI-based acceptance testing, then drive the domain layer with unit tests, including injecting the “random” algorithm to make the tests repeatable. Seemed to go OK, but it’s always hard to tell with these things. As usual, my main aim was to at least show people what a short “red-green-refactor” loop actually looks like, to dispel any misconceptions that TDD means “write all the acceptance tests to match the spec, then keep coding until they all pass”.

I marked Friday’s heatwave by running the Bury Friday 5. Not altogether the most pleasant experience, especially on top of a 40 minute delay getting there thanks to the A14 being closed because apparently it had melted. My slowest time of the series so far, but it turns out each of them has been slower than the last. Surely I can buck that trend at Stowmarket next week?

I failed to finish the hedge cutting that I started last weekend, blaming a combination of heat and lack of brown bin space (although it was mostly laziness), but I did at least spend a bit of time trying to pull up of cut down the worst of the brambles in what I’m stubbornly referring to as my “wildlife lawn”.

I encountered this chap on my long run today. Fortunately he seemed quite chilled and didn’t decide to charge at me, but just being stared at as I passed through his field was a little unnerving.

Leave a Reply