A few weeks (months?) ago I thought it’d be neat to have a little pop-up window visualising the output of jj, so I could look at the changelog graph change in real-time as I did commands, rather than checking Tl (jj log) repeatedly.
En route, I really wanted to know what song was playing on my Inkplate, so I hastily threw together suena and a script for calling it over the network.
As I started learning how to use jj-lib, I decided it was finally time I would not just push it to GitHub, but make my own. nóssa is very much a work-in-progress, but it’s functional (and backed up) enough that I can just keep everything there now. Along the way I learned to produce commit graphs the same way jj does, and get used to git2-rs, which has been neat.
While pushing LiveViews a little too hard, I found I needed another Elixir/Phoenix project to work on, since nóssa’s a first-of-the-kind for me and I needed to learn faster and spread the mistakes out a little. Redoing my blog seemed like a good opportunity. I’ve been putting off posting more until I finally got asset management done, but then I was finding the compile/Nix-realisation-time on my VPS prohibitive, so now I have a really big homelab server (mothdust!) for the first time in forever.
Now I’m thinking I’ll want a caching reverse proxy that sits on a very small VPS, so that if mothdust goes down, GET requests can still work. And maybe I can make it do the upstream requests over the Erlang network instead of just being a dumb client. Among other things, this avoids having to bypass Anubis or similar things another way, and could also e.g. fetch a list of known pages to “pre-render” the site into cache, purge the cache automatically on code updates and content edits, etc.? I’m particularly trying to have fun with my website, because why not? It’s a part of me.