What programming/technical projects have you been working on?

I recently quit both my paying jobs to focus on my side-project full time. It has been a while since I have posted on one of these threads as I haven't had much time to work on the project. For those unaware, my side project is called gametje.com and you can try it out right now! For some additional context, you can check out my previous Tildes post which has a lot more detail. The TLDR is that it is a site with games similar to Jackbox Games. It is playable in 9 different languages without needing to install anything (no Steam or Consoles etc). The idea is create it more like a platform with new games being added on a regular basis.

I've been busy in August working on a major UI overhaul for the main pages and an integration with Discord. The UI overhaul is a result of working with a UX designer earlier this year. The new design is not yet live. The current one is functional but a bit uninspired so don't judge it too much yet :) I'll probably post again next thread when the new design is live. I've also been working on what I am calling "game sessions" which are sort of a lobby before an actual game starts. This allows players to join the session once and switch games without having to enter new game codes. This mainly came about as a result of integrating with Discord as their embedded activities are tied to an instanceId. I needed to give the hosting player the ability to switch/start games from that centralized point. It also needed to be able to figure out which game session to join when clicking the button in Discord itself. Game sessions in general is also a big win as you don't need to fiddle around giving out a new game code every time you want to change games.

I've also been working on adding Discord as an Identity Provider for logging in with OAuth. This is related to the integration but also a bit separate as you can log into the site using the OAuth flow outside of Discord's app. Needed to do a bit of refactoring around the user providers as previously I only had my own username/password provider. This should also make it much easier to add new Idps like Google etc later.

As a part of the Discord integration, I've also been tightening up the websocket code so it can properly detect when clients disconnect and remove them from the game/game session. That's been fun reading through some of that code and playing with it.

Overall, very happy with my progress the last 3 weeks.

Next up:

• Change my hosting provider (probably to Hetzner) for my backend/database.
• Finish the Android TV app to replace the Amazon Fire Store version as the store is being shut down on August 25th and the web-wrapped app it created was never the best experience.

I've been moving back to physical media for the stuff I know I want to rewatch in 10 years. Movies and TV-shows just keep jumping from provider to provider on streaming and some just completely disappear because of corporate bullshit.

There is a local site that sells Blurays and 4k Blurays and has pretty good sales at times, BUT their site is a horseshit mismash of slow-ass javascript and infuriating to browse as they don't give proper filtering options.

So a few days ago I started Claude Code and created a system that uses Python+Playwright to browse through both categories, grab the links to the movie product pages and store the prices to a local sqlite database.

Next up is adding a watchlist + price tracking for each so I can hop on the sales when they happen.

This is my first personal project I might actually release to public use if I can get it polished enough and can be arsed to add some kind of user accounts. ...or just store the watchlist in the URL so that no accounts are needed... 🤔

So I was wondering how to rewrite all the previous git messages in my journal from the time that I was using gitupdate. I mentioned a couple weeks ago that I wrote my own replacement:

With some extra time on my hands I made something a little more descriptive. If you're just modifying one file it will include the edit that you made into the commit message, or if too long, count the number of added, deleted, and moved lines

I had to modify my wip_message.py script to not use --staging only but the script is a lot more flexible now and I did this to rewrite all the messages in my history:

git rebase -i (git log --reverse --pretty=%H | sed -n '2p') --exec '
      git commit --amend -m "$(wip_message.py HEAD)"
'

Pretty cool!

Actually, I should also add some additional flags to also keep the author date original.

Another git related thing that I do, and probably more universally useful than the above, is that I have short commands for working with files based on their git status. For example I have this fish shell function:

function added
    git status --porcelain --short $argv | awk 'substr($0, 1, 1) == "A" && substr($0, 2, 1) == " " {print substr($0, 4)}' | string unescape --style=script
end

then I can run something like open (added) to open all the files that are new. It's really useful! (but only if you remember to use it lol--sometimes I forget).

I revamped these to be more robust and put them all here:

https://gist.github.com/chapmanjacobd/5ecb8be86f83dcfdcd043cf3b68b0b27

Here are a couple useful tools I found this week:

• https://github.com/dlazesz/tr-unicode
• execsnoop-proc, execsnoop.bt (The proc version is older but still compatible with modern Linux kernels; execsnoop.bt is best but relies on having bpftrace installed)

While procrastinating revision, I drafted up a quick REST interface for the thermal printer that my partner grabbed second hand last week, and I've been letting online friends hit the endpoints and print whatever (and we only had one short period of mostly smut).

I'll get back to revising now, but I'm really struggling to force myself to cover the material on Greedy optimality proof by Induction & Contradiction, and Divide&Conquer isn't any more interesting to me. Maybe I'll skip ahead to Network Flow & NP-Completeness for now. Given the exam is a 45 minute oral examination, and considering the amount of material to cover, I'm guessing it won't be too proof-heavy and more general theory..

Moved from ZSH to Fish, mostly because of the sluggish startup times of OMZ, I don't even have that many plugins (2), and following this guide did not improve it much.

Another major contributor to slow startup times was nvm detecting the correct node version on each directory, this was solved by fnm.

The moving was easier than expected, the defaults of fish are better, and having web config is genius, the only issue I had was git completions, I was too used to the ones from omz, and even with this plugin they vary a bit, not sure if I will write my own aliases to match my muscle memory or simply adapt myself to the existing ones to keep it more "zero conf".

Other than that, I got to a point on Rust where I understand all the syntax, ownership, borrowing, lifetimes, etc or at least it does not feel so alien anymore, I have to decide what to build with it yet, my background is mostly in frontend, so I look at Rust like a tool to build UI which is probably not the best use-case, I played with two interesting projects, Dioxus and egui (Their example page is incredible in terms of performance and detailed examples).

I should build something outside my comfort zone, more backend or library oriented (I take ideas), I do prefer to build something of substance than random exercises.

If you need convincing to use Rust, this guy does a wonderful job with his videos: @NoBoilerplate

if you steal audiobooks from libby, they give you the mp3s and a metadata.json with offsets for chapters. i did up a script to split out the mp3’s into chapters (no transcoding) and then i ru it through m4b-tool —- but i’m not sure this is transparent.

all in all, it works well. i liked the old Overdrive system more, but it is what it is. i don’t know what kind of person hates themselves enough to listen to audiobooks in the libby app.

ultimately, i’d like to have it do it all in one go. it won’t be far off, but i’ve got other similar scripts to redo first. when i got a new computer, i decided to do everything from scratch instead of bringing scripts and configs over. i’m happy i did, but there are some quirks with a few things that are beyond my comprehension. :)