What programming/technical projects have you been working on?

I have a small homelab that's constantly growing and evolving. Right now it's a FreeNAS server and a Proxmox VM host, but I have another potential VM host that I'm working on. FreeNAS is mainly to provide storage for various VMs via Samba or NFS - each VM gets a small slice of local storage on the Proxmox SSD, then a larger chunk of storage mounted on /home for applications to use. I discovered that I can't connect via http/https from outside my network despite my ISP leaving them open (or so they claim), so I'm planning on setting up a VPS to act as a reverse proxy.

On the programming side, I started a project to teach myself cryptography. It's the field that got me into computers in the first place, and I've done almost nothing with it for several years now. I'm just working my way through implementations of common hashes and encryption algorithms in Lua, which is adding its own unique challenges. I picked Lua because something about writing code with it just feels good, there's no logical reason behind my choice. Progress is slow and I'm still figuring out MD2 which is relatively simple, but the RFC has been the only guidance I've used so far.

For context, I'm working on an expense tracking and splitting app. The plan is to be a 100% open-source, self-hosted, ad-free alternative to Splitwise. (Written entirely in Rust and Tailwind CSS, by the way.)

I realized Settlemint is a name already taken by some crypto thing, so I'll need a new name. Thank god I resisted the urge to work on branding or buying a domain.

Anyway, the app now supports separate expense groups, so right now you could theoretically use it as an expense tracker, but you'd have no statistics (those are coming, but later!).

I spent some work ensuring my database schema is normalized and coherent (I can finally say my college classes were useful for something, lmao) so I can easily query for, say, expenses made by a specific user in a specific group. I'm concerned about performance (Splitwise is really snappy, so I need to match or even beat that, which will be a challenge) so I expect I will need to flex the design a bit, but we'll see.

Now I want to let multiple people join a group and add expenses, then split those expenses between the group members. I'm not 100% sure on the best way to implement that, but we'll see how it goes.

I also made the UI not suck. It's not pretty, but it's not ugly either, it's quite literally the bare minimum with a little bit of polish. My partner has been pushing a lot for a UI redesign though, which I'm willing to do (provided he actually wants to spend a few days in Figma to design it), but that will happen only after the MVP is ready. Right now I want to have the functionality nailed down, so I can import my Splitwise data (oh god, I'll need to figure that out too) and start dogfooding ASAP.

I started (re)learning Rust. It's been fun, it's been difficult at times, but I like that. Originally I started out with the intention of using gpui to make an apple music client for linux, but after reviewing the API documentation this morning I realized it's most likely impossible. MusicKit is for Swift, and MusicKit.js is for web apps. I don't believe there's any way for me to integrate Apple's js library into a Rust application, if anyone has intel that suggests otherwise I'm all ears!

rejiggering my 3d printer. Im moving all the electronics to an external enclosure, and adding cable chains. RN im modifying my electronics enclosure to fit my specific power supply and control boards, as well as learning how to use buck converters to power a RPI off of 24v. after that, the cable chains, and after that, I need to cut, strip, and crimp over 50 cables.

I've been working on an app for managing Random Intermittent Reinforcement (also called Variable Ratio Reward Schedule) for personal benefit. I posted about it a while back, but now there's just a little bit of red tape left before getting something out to other users. I'm really excited for the release because figuring out how to use this concept in an active, intentional way has had a lot of impact on my life.

On the technical side, I've been figuring out how to increase the back-end's resistance to scripting so I don't have to collect user information as early in the on-boarding process. I always hate when I install a new app, and instantly hit a wall where I have to fill out forms and check emails and enter codes in order to do anything. By contrast, in my current prototype, the app is usable without an account and a user will unlock a good amount of features before deciding if they want to create an account which also does not require an email/phone-number. The app will need some form of identification for account recovery most likely, but that can come later when it is not blocking the user from doing basic things. All the niceties for the user does mean I have to spend some time to make the API a little less easy to script (but also to debug 😢), but I'm betting it pays off.

If you want to take a look at the landing page and get notified when it goes live, you can take a peak at https://www.tantaluspath.com/ and fill out the relevant google form. I'm always looking for feedback, so I hope if you see something improvable you let me know!

I've been creating a tacacs server as part of my NMS platform for network devices with authentication happening against Azure Entra. User SSHes to a switch and gets a URL and code, user goes to URL and enters code and logs in to Microsoft Entra. User presses "Enter" on router SSH prompt and gets in with list of commands/priv associated. Password never goes through the network switch =D but validates the user authns through MS. In addition, I now collect accounting information from said devices, normalize it and ship it off to Azure log workspace table (and trigger auto backups/diffs). All of this along with authorizations gets controlled through the web UI. Next part I need to do is setup token forwarding (or similar) from the web front end to the switch as the 'password' then validate the short lived token is valid, that way a user logs into my NMS UI and clicks a device and automatically gets SSOed against their Azure creds and straight to the switch. Obv, there's a lot more to this like offline 'break glass' accounts, caching, etc, but it's coming along! Very satisfying feeling to see the pieces come together and work with real Cisco, Fortinet, Aruba, etc devices after so much effort.

I started writing my own property-based testing framework using Typescript and I think I'm making good progress. The API's are looking good. The next thing will be to start on the shrinker.

Working with Arbitraries is looking quite similar to logic programming. I should write an n-queens solver demo.

welp, I decided to take my little calendar thing a step further.

Here's how it was:

1. google sheet with date in one column and the event title etc in another
2. in another column I've got a formula to generate the .ics file
3. I publish the spreadsheet as a .tsv and limit it to the range of the generated .ics file. the tsv is only so it'll be plaintext. There won't be any tabs separating anything.
4. on my VPS I set a cron job to wget -O calendar.ics https://docs.googe...
5. I subscribe to the output

Not the worst system, but it would be nice to not need the VPS involved. I decided to go with github actions...

name: Calendar - Television

on:
  schedule:
    - cron: '* */2 * * *'
  workflow_dispatch: 

env:
  FILE_NAME: tv
  SHEET_URL: https://docs.google.com/spreadsheets/d/e/2PACX-..._/pub?gid=1234567890&single=true&output=tsv&range=G1:G

jobs:
  download-and-commit:
    runs-on: ubuntu-latest

    steps:
    - name: Checkout repository
      uses: actions/checkout@v4

    - name: Download file
      run: wget -O ${{ env.FILE_NAME }}.ics --accept tsv "${{ env.SHEET_URL }}"

    - name: Configure Git
      run: |
        git config --global user.name 'github-actions[bot]'
        git config --global user.email 'github-actions[bot]@users.noreply.github.com'
        
    - name: Check for changes
      id: changes
      run: |
        git add ${{ env.FILE_NAME }}.ics
        git diff-index --quiet HEAD || echo "has_changes=true" >> $GITHUB_ENV

    - name: Commit changes
      if: env.has_changes == 'true'
      run: |
        git commit -m 'Add downloaded file'
        git push origin main
      env:
        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Runs every other hour. Update the sheet and the calendars follow suit. ezpz.

Working on a personal project for a MUD I play, basically taking color codes that are in the MUD and displaying the correct colors in browser using spans.

Everything is working correctly. I mostly wanted to play with Astro and islands, which has been fun, and now I'm just trying to make things look pretty.

I picked up an e-ink panel and I'm playing with making it show home management information ("it's trash night, take the trash cans out to the curb", "laundry is done", "you left the garage door open", a calendar, maybe package tracking?). I've been referring to it as trying to automate away our executive dysfunction.

The Home Assistant API is... kind of asinine. Let's say you have some devices with battery level sensors, reporting that data to HA, and you want to show an "X has a low battery" alert for each one. So you go and ask it for a list of entity states, find the ones with class "battery" and id starting with "binary_sensor.", and then... uh. You can't get the device name from there. There literally isn't an endpoint that returns device information (an "entity" is an individual attribute of a device, devices are the physical objects and have multiple entities each; these are not the names I would have chosen). The friendly_name of the entity is something like "Device Name: Low battery level" by default, so you can parse it out of that as long as nobody renames anything. Or you can use /api/template to ask the HA server to render a string template containing the value you need, because that has an entity-id-to-device-id function.

I've been playing with some geospatial AI stuff recently, since I did a training course on it a month or so back. I've got a lot of geospatial experience (I work in the field), and quite a bit of AI knowledge but hadn't really put them together much before.

I spent a while putting together a simple app that can search an aerial photo using text queries like "roundabout" or "school playground". I got it working on my computer a while back, but working out how to deploy it properly on the internet, with a server that was powerful enough to deal with the AI model, HTTPS and all that jazz took a while.

The current version of it is working at https://server1.rtwilson.com/aerial - feel free to have a play, comments welcome. I definitely have some things to improve for the next version, but it's a reasonable tech demo at the moment, I think.

I started looking at migrating an web app written in asp.net with ef core to use postgres instead of sql server. I quickly ran into a weird issue that updates failed (expected 1 row affected but 0 rows affected). Still haven't figured out why. I thought ef core would be largely agnostic when it came to this type of basi operation! The updates essentially maps a domain object to an entity, then calls something like:

var entry = _context.Entry(entity);
if(entry != null) 
{
    entry.CurrentValues.SetValues(entity);
    entry.State = EntityState.Modified;
}

await _context.SaveChangesAsync();

It's weird to me that the updates fail now, but worked when the db was sqlserver... have to look over the migrations to see if there's something weird there that is causing issues... inserts seem to work 😕 🤷