What programming/technical projects have you been working on?

I keep thinking about re-building my blog. My last blog post was about building a bespoke blog website using Bun, TypeScript and Hono. I am starting to think of how I want some features like: pages, search, tags, images, etc. That want is making me think about going back to 11ty or learning Hugo. With all this indecision I have also not been able to write anything for my blog and that is the goal of a blog, I think.

In the few minutes of spare time I have between my job, a fresh baby and some projects I have mentioned here in the past, I started this week looking at some of the tildes source code.

I'm considering if I want to put in the effort of building (part of) the JSON api for tildes. As far as I can tell, this was discussed a couple times in the past but never started. It honestly seems pretty doable simply from a coding POV.

But I have a feeling it will involve so much debate on how to do it 'properly' or on which library to use for it that it is making me second guess myself before even starting.

Regardless, just getting caught up on the source code of a platform I use a lot is a fun activity in its own right. So I'll just do that for now

I subscribed to Claude Code Pro for $20 and have been knocking off my todo-list of "this might be useful but I can't be arsed to write the code" -apps.

Just last night I built a Bubbletea TUI wrapper for yt-dlp with queueing, support for playlists and download directory selection. Took me 2 hours, of which maybe 10-15 minutes was hashing out the specs and libraries with gemini-cli (gemini-2.5-pro) before I put Claude Code on the task. Claude wrote the code on the second display while I watched some Anime on my primary one.

Had to poke it a few times and remove a feature (output log display) that was messing everything up, but now it just works. I can just switch to the tab it's running on Wezterm, paste url and go about my day. No need to worry about urls with ?, # or other CLI-special characters. Nor do I have to wait for one to complete before adding another.

So I've been working on Storyteller v2 for about 4 months or so. It's a huge update, partly because I want to get all of the potential breaking changes to the API in there, and partly because I took the opportunity to completely redo the database and auth layers. Since I started, I've changed.. a lot:

$ git diff --shortstat 35f1b19a..main -- . ':!yarn.lock' ':!**/__fixtures__/**/*'
 267 files changed, 14416 insertions(+), 4998 deletions(-)

Out of a total of:

# I removed a bunch of `-not -path` args to make this easier to read
$ find web -type f -exec wc -l {} + | awk 'END{files=NR-1; print files " files, " $1 " total lines"}'
297 files, 36308 total lines

So I've touched 90% of the files, and I've changed at least half of the individual lines of code. 😮‍💨. I'm really close to being ready for a beta release, though, and that's exciting!

Also I made some very nice little animations in various places, and I'm really happy with that. I'm used to working on internal tools, where "delightful" animations never have priority (even though they often can do a lot of work to make a UI more legible!), so this is the first time I've really worked on CSS animations, especially keyframe animations. What a nice API! CSS is incredible.

Finally took the time to write the measurement tools for my VTT, such as a basic ruler, cones, circles, etc, that I will use to finish up the drawing tools soon as well. I also sat down and added some functionality purely for comfort, such das dragging and dropping tokens and maps between folders and other library related features.

We are having a blast using my VTT for the past few weeks.
By now other GMs of our group are using it as well.

Hey y'all, I haven't posted in a while. I'm still working on my game. Progress has been very slow these past few months, and a lot of the work that went into it are very... unglamorous? It's mostly about progression (how the player gets stronger) through a tile-based dungeon building system.

The way it works is:

1. The player picks a tile to build a room
2. The player picks a room to build
3. Some rooms require the player to pick a monster tribe to live in it, in which case, that happens in this step
4. The room is built, and depending on what it is, the player can do a few things - upgrade their stats, acquire abilities, acquire effects, re-roll available options, etc.

To make this more interesting, I spent a large amount of time working on the monster tribe mechanics. Each monster tribe:

• Offers a different set of upgrades/items, so the monster that lives in a room determines what options the player will have.
• Affects the atmosphere in the surrounding tile - the temperature, heat, and noise level in the surrounding tiles of a room that a monster moves into.
• Has likes/dislikes when it comes to the atmosphere of a tile - if there's something about the atmosphere they like, the player gets a bonus. On the flipside, if there's something about the atmosphere they don't like, the player gets a penalty.

With each room that gets built, and each time the player does something in a room, enemies on the map will advance one tile towards the player's "boss room" (you play as the boss monster of a dungeon). Once an enemy reaches the boss room, the dungeon building phase ends and a combat encounter begins (where all the upgrades you've been working on will come into play).

I feel like this took me way to long to do... I started working on the monster tribe stuff around the end of March of this year. Finally, as of yesterday, I finally have a prototype build again that I can ask people to play.

I wanna start working on some more enemy variants - the output would be a lot more shareable (gif-able), and I'm expecting that it'll be more fun that working with mostly UI and systems!

Per chance anyone has any advice on this, although this is an unusual spot to ask for this.

A fresh, new Arduino that I ordered arrived a few weeks ago. And unfortunately I couldn't really start because of a lack of headspace. Now I have some more room in my mind and I want to start but can't seem to start? Closest thing I've gotten to with describing it is a bit like an artist block, though I'm not sure if it's the same since well, I'm not very artsy.

Does anyone know how to deal with it? Either on when to try to start, or how to deal with the frustration that comes with it?

I recently became interested in hosting something on Oracle's always free compute instances. For this project I have a few design constraints:

1. Everything, except secrets, are self-contained. This project probably would've been reasonably well suited to a docker image, but I didn't want to deal with the hassle of storing the image somewhere so it's all just in the cloud-init.
2. The application should have it's traffic route through a VPN. This largely just has to do with the nature of the traffic.
3. The application needs to have a publicly accessible endpoint. Due to 2, this means I need to tunnel out of the VPN.

Overall, the structure is simple: install deps, fetch secrets, run VPN, run application, run tunnel. Oracle has made this more of a headache than it needs to be though. Their dashboards are incredibly confusingly designed and very slow. Some of their tools also behave weirdly.

Maybe my biggest headache was oci CLI. For those unfamiliar, oci CLI is Oracle's equivalent of Amazon's aws CLI. I thought that it would be trivially easy to install into Oracle's own Ubuntu images, maybe even already being installed. Nope, it does not install nicely at all in a cloud init script. It is based on an interactive shell install and if you put it into the headless mode that accepts all defaults it throws a permissions error. Pretty cool that an image as popular as Ubuntu doesn't even cleanly install the CLI that you use to fetch secrets from the secret vault. I ended up using the containerized version of the CLI, which feels completely absurd to me since it's to get an Oracle CLI onto an image from Oracle's own curated list.

And don't get me started on their ipv6 mess. I'll just leave it at that even their auth servers are incapable of servicing ipv6 traffic, so hope you don't want to fetch any Oracle resources from inside your Oracle instance unless you have a public ipv4 address because you won't be able to authenticate.