What programming/technical projects have you been working on?

My work-in-progress “cozy community site” plots.club has a pilot site! I’m currently looking for beta testers for the site - nothing heavy, just looking for feedback from real people. I have a lot more information in my latest devlog post about it if anyone’s interested in joining the experience! (You can also just PM me here if you’d like). Hope the self promotion of the site is okay, I’m not actually making money off it!

I upgraded my 3D printer. It was a Prusa Mk3s and I upgraded to be a 3.5s. The upgrade wasn’t easy - I made a couple of blunders that required backtracking. Now that it’s done, though, the increased speed is welcome and the print queue in Prusa Connect certainly makes it easier to keep it busy printing parts. I’m getting tangled filament more often, perhaps because I’m not going out to the garage to check if it’s done yet, so I don’t see it before it fails.

Unfortunately, Prusa Connect, while nice in other ways, is very bad at delivering notifications. The Android app doesn’t seem to deliver them at all when the phone is asleep, and then I get a bunch of stale notifications later when I actually look at my phone. Getting notified when a print job is done is the thing I most hoped for from WiFi connectivity and the mobile app, so it’s disappointing in a ‘you had one job” kind of way. However, if I keep the Prusa Connect web page open in a browser tab, the percent complete appears in the title and I can see progress at a glance, even while doing something else. It’s still an improvement on running out to the garage.

I might try Prusalink (the non-cloud version) next. Maybe it’s all I need?

On my musical keyboard project, I increased its size to one octave (with no electronics), which required increasing the strength of long and thin parts that started bending. I’ll work on scaling up the sensors next.

I'm currently experimenting with using a locally hosted LLM (llama.cpp + gpt-oss) to translate and reformat some course material slides that weren't provided in English, and if I manage to get a RAG pipeline set-up, I hope to get it producing formatted markdown which contains the slide contents cross-referenced with the book that the course is based on. Yes, I'm doing this instead of revising said material because I physically cannot force myself to concentrate for long enough to read said book, which is almost nine hundred pages of dry academic text on Algorithm and Datastructure theory.

I am trying to figure out a new work-from-home set up. I just built a new PC and I'm trying very hard to not entangle my personal PC with my work. I think my monitor has a KVM switch but I can't figure out how to configure it (even with MSI's provided youtube guide). I thought I might be able to just add an additional monitor to my desk but the new PC is around 60lbs (27kg) and the monitor itself is another 20lbs (9.6kgs) and I'm worried about my desk collapsing with the addition of my laptop and another monitor...How do y'all handle all this weight on your desks?!

Storyteller has had 107 pre-releases for v2 😩. We're getting there! Very, very hopeful that I'll have a real pubic beta release next week.

I'm writing a software mono synthesizer. The idea is to eventually incorporate it and a drum synth in a self-contained software groovebox and hopefully run it on one of the really cheap Chinese Linux Game Boys you can find on Ali Express. The synth is based on Casio's invention of phase distortion synthesis, but I try to put package it more like a Roland TB-303, with accents and portamento, and with few knobs and controls yet many sweet spots, but also with complete MIDI control.

An interesting, bite-sized problem so far was implementing voice prioritization. The key you pressed last always has the highest priority, but if you release that, the one you pressed before that should have priority and so on, and you can of course release the keys in any order.

I opted to use an array with one entry for each of the 128 possible MIDI note pitches. Each entry represents a doubly linked list node, pointing to the previous and next pitches, if any. So on one hand I can link them into a list where the tail is always the highest priority note, but when I unlink them when a note off is received, instead of iterating over the list, I can simply index the array by the pitch to retrieve the node. In that way, there is very little pointer chasing and no iteration over the list at all, and the nodes are all contained in a contiguous 768-byte array, so the common caveats about linked lists (locality, O(n) iteration) have been eliminated.

One open problem is designing mapping curves, in order to map the user controls to sometimes several internal values. I feel like a visual tool for designing curve functions where you can just drag points around would be helpful here, but I don't know of anything off-the-shelf, so if anyone has an idea I'm happy to hear it before my project becomes two projects. I have curves that work for now, anyway.

I used to store my bookmarks on pinboard.in, but stopped paying a few years ago.
It turns out pinboard has an export option, so I've been working on a clone for my personal use, and just imported all of my old bookmarks.

Really, the project is just an excuse to have some fun and experiment with some stuff I have an interest in (web frontend with HTMX for interactivity, and Erlang backend).

Recently started into 3D printing and been thinking of things I want to print and getting used to the different settings when sliced (infill/etc) and how they work and getting ideas of how much filament and time each kind of print i want to do would take, etc.

I suppose it's mostly "technical" in the aspect that I'm learning about it.

Mostly want to print personal stuff but I have considered that maybe I could slowly start a business making 3D printed things? Not sure

It sure is handy to feel like, when I need a little bit or bob to serve a small but important purpose, I can just simply print those now. Especially things like stands for light things, covers/caps, handles/holders, etc.

I've been tweaking my disk-usage script.

I noticed that my published version isn't counting subfolders correctly and the immediate children vs recursive statistics are not clear cut. It's kind of a mess.

Google Gemini has been teaching me prefix trees (trie) but I'm not totally convinced. Counting subfolders is more accurate now but I still feel kinda bleh about it. I'm thinking of starting over again... but it is mostly working. I think I'll start over and not worry so much about doing too much within each loop and just make a big quadratic time loop but it will be simpler and easier to understand

At work I'm basically redoing a lot involving registrations of time(C#/.NET). Having an understanding of measuring time as a physicist is very helpful because I'm used to think in a different way.

But it's still hell. You have to be so careful about what you do and make sure you don't implicitly add more information about the timezone or when it was measured. Even things like schedules contain more assumptions than you'd think.

I'm in the testing phase of the Stremio plugin I was working on. Some parts of it are a bit hacky, but it is overall working pretty well. I'll probably be able to release it soon even though some stuff it does isn't exactly ideal.

I've also now entered the setup phase for my other project: a tunneling service. I tried both cloudflare tunnels and local tunnel, but neither of those nor any other options I could find really satisfied what I wanted. Most notably local tunnel lacks encryption and ability to reserve sticky names while cloudflare mandates having a custom domain and the clients kind of suck (don't really function correctly when trying to tunnel out of VPNs, have issues when not root/administrator, etc.).

So the other week I posted that I picked up a locked remarkable 2 tablet for a song, and sent out for supplies to unlock it. They arrived.

Suffice to say, any hope I had of being a surgeon are dashed. I'm crushed. I'm amazed I can function in society when it became how apparent my lack of hand steadiness is. I persevered, and managed to solder the pogo pins to the micro usb breakout, and by some miracle, it actually came out pretty even/straight. And I only singed the pcb a little. Thankfully, I opted for a usb c to dupont breakout so I don't have to solder more things.

Proceed to the actual unlocking. Of course, I didn't consider how I was going to get the makeshift pogo connector lined up and making good contact. Thinking quickly, I used a gift card, a wad of paper, an old textbook, and my hand to hold it in place [0]. It worked pretty reliably, I actually expected the pogo connector to give me the most problems, but it's actually the dupont jig that became the real headache.

You see, it requires shorting b8 to b12 with a 10K resistor, but as you know, usb c can be plugged in 2 ways. Those wise in the ways of usb will be quick to point out that that's a trivial issue. The real issue is if the cable even supports those pins.... and I have no idea. After spending a morning trying various cables in various orientations I'm left with two issues, which may both be true: the breakout pcb is problematic/mislabeled (which seems to be a theme in the github issues); the usb c cables I'm using suck (which is also a theme in the github issues).

I'll send out for resupply and mount my assault again when I'm better equipped. Torn between getting a proper cable, or just buying a male breakout and skipping the cable all together. Probably the latter. Having one cable in the mix should make debugging easier. Maybe both.

[0] Surprisingly, the springs are strong enough that the textbook wasn't enough to keep it from moving, hence the need for some human input. The gift card + paper wad help align the elevation, while the text book keeps the pogo from popping upwards, as well as acts as a mass damper to suppress the vibration from my overly caffeinated self.