Weeknote 17: Reverse dry January

It's Febby! It's also the day after first payday (for me) so I'm feeling quite serene.

What's he been doing then?

Release refinement

A lot of my dev time was spent refining our automated release setup and boxing with lingering problems.

A fun one at the start of the week was making tests run on pull requests generated by our workflows. We have a GitHub action that's used to build the release ahead of publishing it, for which we create a pull request from within that GitHub action to check the build before proceeding with the rest of the release process. The problem was that our automated tests refused to run on this generated pull request.

After going around the houses it turns out this is an intentional quirk by GitHub that anything generated by a workflow doesn't trigger other workflows so that you avoid an endless loop of workflows triggering each other. Peter Evens, the maintainer of the Create Pull Request GitHub action, has already documented this which is very helpful. We went for the PAT solution if you're interested.

The other thing I started looking into just yesterday was better supporting pre-release syntax, a lesser known feature of the SemVer spec. It would make it easier for us to release things like beta versions or internal versions for testing. So far it's just a lot of string manipulation and regex but it's an excuse to clean up the code.

We tested our work with an internal release yesterday and, whist there are few lingering issues, it's looking really nice. It's way way faster than our previous process and a lot less boring. I've overall found this work really rewarding. A nice tidy code project that also acts as an opportunity for me to get my head around GitHub actions. Perf.

A little critique of GitHub's docs

Throughout this work I've spent quite a lot of time trying to get info out of GitHub's own technical documentation and it frankly isn't structured in a way that works for me. This has been consistent with any time I've had to interact with GitHub on a technical level such as trying to work with Octokit.

The GitHub actions documentation for example spends a lot of time talking about the technical concepts that underpin a given action option. I understand why this is useful and it's good that this is written down, but truly all I want is a list of possible options for the on: parameter. It's too far the other way where the conceptual writing has erased the summary for the layman.

This feeling is a shared one from talking to technical colleagues, so something's going on. I've got some hypotheses about GitHub's internals such as how powerful developers are within the organisation and who they're recruiting for research but that's just speculation. Either way, being on a team that I think does a lot of work to make their documentation as economic and digestible as possible, I wish other organisations would bother to do the same.

The last week of Jan

A comparatively quiet week outside of the job. Did a little bit of hobnobbing with folks at the pub on Thursday for Tim Paul's "leaving" drinks. Good luck Tim with being the head of design for i.AI, although with the new GDS you're still going to be our colleague it turns out.

I finally started on the new Zelda where you actually play as Zelda, a game I bought several months ago and has been sitting in it's plastic wrapping until this week. I really like what it's doing. It's a challenge to some of the mindlessness of classic Zelda games where you just mash the attack button and get on. Now you're not as powerful you have to think about stealth, stuff you can use in your environment, summoning AI (steady on) monsters for you. How do you manage them? What can you do to help them? It all feels very fresh.

As a general reflection I feel like a lot of the wellbeing efforts I put in place in the last leg of 2024 have fallen off this month. I've had a bit of a reverse dry January. A literal example of that is how much I put into reducing my drinking in the latter half of last year only for my sobriety to hit a consistent low over Jan. It's a bit trite in my opinion to use months as start and end points for behaviors but with Jan over, I'm hoping I can get a better grasp on my physical and mental fortitude.