What programming/technical projects have you been working on?

For the past few years of Advent of Code, I've written step-by-step explanations for how each puzzle is solved. This has been an important step in my learning, plus it's fun.

Previously, these writeups have been available as GitHub READMEs, which is fine. But, in an effort to really showcase them, I'm making a little static site to host them. This has a few advantages:

• I can get a little fancier with my syntax highlighting (e.g. line numbers or highlighting new rows)
• They'll be easier to find rather than buried in a repo
• I can add things like an RSS feed

I'm building it with Astro and am very impressed so far!

I'm making my own frontend Javascript framework (happily adding on to the many that already exist). I've been in the web development world long enough that I feel I can make something better for my small side-projects, which I frequently make. It's been fun thinking from ground zero, deciding on what to keep from vanilla Javascript vs what to abstract away. One thing I find enjoyable about it is that I can limit its capabilities and reduce upfront work because I'm making it for myself. My basic strategy has been to build in new features when the need arises while developing a side-project.

I had an idea tunnel into me that involves some graphics shader hacking/mistreatment. After a week of moments when I was sure it wasn't possible, and refactoring the architecture 3 times, I actually generated realistic, usable output.

A long way to go still, but it's hard to describe the relief and satisfaction I felt after everything fell in place. And all of it is a little too technical to try to explain to people in my circles.

I am currently preparing to work on KDE Plasma 6, be it code or just testing. Also since I have now time thanks to being laid off (..yay..) I think I finally have energy to work on my game project.

For the past year I've been working really hard on Storyteller, and I finally have it more or less working end to end, which feels amazing.

Storyteller is meant to be a self-hosted, open source replacement for Amazon Kindle's Whispersync. I love the ability to seamlessly switch between audio and text when reading a book; I prefer reading text but very often find myself wanting to continue my book when I get into the car for a long drive. I also really don't like using Amazon products, nor DRM-protected content, nor cloud services for media content.

So now I have Storyteller! It's got three pieces; a backend API (that does library management and automated text <-> audio synchronization), a web interface for administrating a library, and a (not yet publish) mobile app(s) for reading and listening to the books.

I've gotten to work with a very enjoyable mix of new and familiar technologies for me. The web interface is Next.js, which I'm very experienced with, but the mobile apps are React Native which is brand new (and has been super enjoyable, with some notable pain points), and the backend API is using fastapi and whisper, which are both totally new to me. If anyone wants to chat more about these, I'd be happy to.

This whole thing is definitely still in alpha, but I'm hoping to finish up the last few features (collections and user invites) by the end of this year and "launch" (basically just make a documentation site and announce that it exists) a beta early next year.

My 3D printer has some long overdue upgrades and enhansements that need to be done.

I've already put together a Nevermore filter for it, and now I just need to actually install it on the thing. There's also the issue of actually procuring the activated charcoal for it. When I installed the acrylic panels to enclose the printer, I somehow missed that I was supposed to apply some felt tape so it would seal properly. I might just skip that though, as I opted to use hammerhead style t-bolts to attach them and they are a real pain to reattach.

The super floppy hinge on the case for the control on it has always annoyed me and there has been a vastly improved version available for probably the past year. I've got it printed and I'm just waiting for a missing screw to come in the mail.

Along with that screw order I also got all the screws I would need to put together a stealthburner toolhead. The afterburner has been enough for me for quite a long time because I found that ABS is basically perfect for everything I print. But for the first time in forever I find myself needing to print something flexable, and much to my chagrin it's not able to handle TPU. I have a roll of PP that I would also like to try to print with but it's having the exact same problem because it's so flexible. I haven't bought the other components needed for the stealthburner yet, which is something of a problem since it will probably take a long time for it to come in.

I'm the main dev of Omnisearch, an Obsidian plugin. Nowadays the project is in "active maintenance", but I got a burst of inspiration 3 days ago and managed to combine a small http server (bundled in the plugin) with a userscript, to surface Omnisearch results directly in Kagi.

A small demo of the proof of concept.

I loved spending a few hours on this simple feature, because I got to work on a userscript. They're underrated and feel like relics of the old web, but they embody the idea of hacking, bending a tool – in this case, your browser – to your will with just a few lines of code, making it a little more yours without resorting to curated plugins or settings.

RSS Reader:. I finally redid the ingestion system to put feed id’s on separate feeds. I did a sample of 10% of the feeds and did a complete ingestion and classification cycle and seemed OK, I switched all feeds to the new endpoint, and am hopeful the next cycle (maybe tomorrow) is successful. I still need to update the UI for managing feeds so it creates new style feeds but that doesn’t seem difficult at all.

One application of this is feed-specific processing. One annoying problem is that anything I submit to Tildes gets picked up from the Tildes RSS feed and of course the classifier will think I liked it because I liked it before! Things like this add up to maybe 5% of my feed and this info will help me filter out items I know I will never like that the classifier struggles with. I’d have to work pretty hard to get a similar improvement with the classifier because the classifier is already strong, particularly given that I slowed down the classification cycle, letting it be more selective and operate at a better spot on

https://en.wikipedia.org/wiki/Receiver_operating_characteristic

Another big feature on the horizon is something that looks at an article from a place like phys.org, extracts links to their sources, determines if the papers are open access, and gives me a choice of which article to submit. But really I should spend time on…

My blog: Still no progress, but a high priority.

Three-sided cards:. No technical work this week, but I am thinking about a simulation system for collages I want to make because I made a huge number of perfectly good prints that will be relegated to work prints

https://mastodon.social/@UP8/111219354336635872

for instance, where after having the cards in my hands it was so clear that if the characters are sharing the same space they have to be scaled appropriately by height the way Rin Tohsaka and Archer are. I am planning larger and more complex collages and my current process of “printing 50% more cards than I need and put them together by hand” won’t scale.

Really though I gotta focus on getting the blog up.

I've been working on a Minecraft mod that adds a procedurally generated solar system that you can see in the sky! Everything has a simulated orbit using Kepler's laws, which was a bit hard to wrap my head around when writing the code, but it seems functional now. Next step is to actually add the procedural generation and do a bunch of pixel art.

I'm making a board game framework with my brother. Hoping to release it soon but so far it's really great, easy to work with and looks solid.

I just added an Exploratory Data Analysis (EDA) subcommand to my desktop productivity tool

pip install xklb
library eda my.csv fancy.parquet pickle.pkl excel.xls jason.json html_tables.html etc.nc

https://github.com/chapmanjacobd/library/commit/1622affba03ef6acad737b0d63b2291dd5d91ac7

Ideally, it is able to read any file that pandas supports (there are some I know I still haven't added yet) but it works with most things. Spits out summary statistics (in Markdown!) like this:

## tests/data/test.csv:0
### Shape
(2, 4)

### Sample of rows
|    |   Unnamed: 0 |   A |   B |   C |
|----|--------------|-----|-----|-----|
|  0 |            0 |   1 |   3 |   5 |
|  1 |            1 |   2 |   4 |   6 |

### Summary statistics
|       |   Unnamed: 0 |        A |        B |        C |
|-------|--------------|----------|----------|----------|
| count |     2        | 2        | 2        | 2        |
| mean  |     0.5      | 1.5      | 3.5      | 5.5      |
| std   |     0.707107 | 0.707107 | 0.707107 | 0.707107 |
| min   |     0        | 1        | 3        | 5        |
| 25%   |     0.25     | 1.25     | 3.25     | 5.25     |
| 50%   |     0.5      | 1.5      | 3.5      | 5.5      |
| 75%   |     0.75     | 1.75     | 3.75     | 5.75     |
| max   |     1        | 2        | 4        | 6        |

### Pandas columns with 'converted' dtypes
| column     | original_dtype   | converted_dtype   |
|------------|------------------|-------------------|
| Unnamed: 0 | int64            | Int64             |
| A          | int64            | Int64             |
| B          | int64            | Int64             |
| C          | int64            | Int64             |

### Missing values
0 NaNs (0.0%)

Let me know if anyone has any ideas for other interesting stats I could pull out