What programming/technical projects have you been working on?

I am loving exploring the Inform7/TADS/Ink/Twine world of interactive fiction. Right now, I'm neck deep in Inform 7 and rediscovering so many amazing games from my youth. I know the first dozen+ things I make will be rough, but I'm so excited to work on this in the evenings.

Other than that, I've been playing with building a novel production workflow out of Vim/Git/Make/pandoc, etc.

Still in the embryonic stages of that, but it's certainly exciting!

I’ve been continuing work on my goodreads clone. This week, I’ve been writing a bunch of html and css to work on the mobile interface for the navbar and the user entry options - the current struggle I’ve just overcome is just styling out a <details>
in a way I think doesn’t look awful, next up I need to work on the dialog forms for adding new entries, then I can work on the backend POST requests those forms will make

I've been building an open source podcast editor that transcribes the source audio and lets you edit it as text. Last weekend, I finished implementing collaborative editing, a 6 week project!

This weekend I'm reworking the transcription queue to run on GPU PaaS platforms like runpod.io, so I can stop running the GPU workloads on my gaming machine 😁

I'm working on a VS Code plugin for doing AI chats in a plain text or markdown file (and probably other text files; I'm particularly interested in Jupyter and other notebooks). The idea is that the chat history will be easier to save, edit, and reuse than doing it in OpenAI's UI.

This is built on Simon Willison's llm command-line tool, which can be customized via plugins to use whichever llm you like, including the ones you can run locally. I mostly plan on using it with GPT-4, though.

I'm not sure if I'll publish it, though. I thought GitHub Copilot only did autocomplete, but it's more versatile than I thought. I need to play around with it more to see what it can do.

I recently finished a python program that monitors folders for new container files, logs them to a database, and then submits the file to the right third party service. Then the files are monitored and when processed they get downloaded to pre defined categories that serve as folders.

I did this before I realized pyloader existed and I could have just built on it.

I did it to replace another python program that stopped being developed and had issues with various python components being the wrong version.

I just made the same thing without integrating with pyloader and made it leaner while expanding on the features I needed and it's cli based.

Now I'm working on an azure runbook to inject scripts into VMs but it's deceptively easy. The instructions are pretty easy to understand but the devil is in the fine print. Bypassing the need to have local storage for scripts makes it very difficult. Microsoft didn't intend for you to run scripts in any other way than a physical file which can't be stored in the vm sandbox that does the pushing.

Next on my hitlist is a C++ application to search for, pull and organize wallpapers depending on sizes into folders via api calls to the wallpaper sites.

I just finished a suite of scripts that theoretically "beats a level" of the game Bitburner with one command. Now I just have to let it run for a while to see if it works and where to optimize it. Then maybe I can stop procrastinating the public release of the tildes keyboard navigation userscript I adapted/made if there's any interest in it.

Nothing interesting, but I’m just finishing up my first complete project! It’s a text compression CLI tool written in Go using the Huffman coding algorithm. I got a little ahead of myself (still have to finish up parts of the documentation) but I published it and made it available on GitHub for people to install.

I am a little worried it’s lacking in options right now but I’m thinking of adding an analytics flag (to tell you how much a file was compressed). Scratching my head for other ideas. Anyway, that’s the project :)

The result of me using the OS and applications on my laptop with a fruit on it more and more exclusively through the Terminal, and significantly less with the mouse/trackpad than what’s probably average, is that I’ve been getting into (bash/where needed POSIX sh, or to be precise more commonly zsh) scripting recently as part of setting up my dotfiles with everything relevant to my current command-line setup. Which also included determining what all standalone scripts are still relevant and making the necessary config files part of a “backup” (git repository that’s also stored and available online), and began with writing a setup/software installation script that pulls/bootstraps everything from there.

So now I wonder: Anybody here got hands-on experience with NixOS? As far as I understood it at least, the “reproducibility” of an environment seemed very intriguing, not to say attractive, to me last that I checked it out.

But there remain two unknowns: How much effort is setting this completely up to reproduce the OS, preferences and installations like I’d want it to be; and how user-friendly is using the OS/setup itself as a daily? Admittedly I haven’t looked into it all that much so far, I was planning to do that this weekend, given enough free time…

I am back at work for two weeks before taking another week off to watch the baby while our daycare provider is on vacation. That being said, a lot of things at work are just taking over our DevOps and tuning a bunch of PHP-FPM/NGiNX containers (ECS on Fargate, if you're curious) for high Magento 2 volume. A lot of it has been load testing with K6 and changing up the Static Children parameters for FPM. Any advice folks have for me here would be hugely appreciated.

I am also considering working on refactoring an HTTP service we need for 2024 to be up and robust in my free time.

I’ve been working on redoing my Webdev portfolio, I made my original one right after learning to use express and bootstrap, so its performance isn’t that great and doesn’t look that good either. Also, all my data was hard coded inside a json file, so I would need to modify the source code and redeploy every time I want to update it. The new one uses a cms instead, so it should be much more easier to update over time. And then it’s back to job apps. Sigh.

Investigating, documenting and testing the enterprise-scale deployments of each defender for cloud plan in Azure. Including how to set up supporting identities, agents, policies and logs from the various resources to allow security analysts to properly respond to any alerts the defenders generate.

It's all fun and games as long as you can use the one-size-fits-all autodeployments, but things get really obscure and not-documented as soon as you require more control over what is happening

Oh, and putting all of this in reuseable Terraform modules for good measure of course!

(1) I hooked my RSS reader to my old autoposter for Hacker News so I can select articles that I like and queue them to be submitted. When I post links to Mastodon I try to always add an emoji at the front

https://mastodon.social/@UP8

I used to cut and paste those from emojipedia, but now I have a dropdown that shows all the emoji I've already used which saves me from dealing with that janky site. There is a lot more I can do on that system but I think the next big priority is adding a facility to import an arbitrary web page so it works as a bookmark manager. I need to also tag items based on the feed they came from but that involves rebuilding the ingestion system and it seems like a hassle.

(2) I've spent about a decade thinking about the semantic web, things didn't really pay off until last year when I figured out a correct mapping between (most) XSL schemas and conforming documents to OWL schemas and instances accordingly. I've got to turn this into an open source product so I have some visible deliverable for that time in my life. I just brought my prototype out of mothballs and need to figure out where to go with it.

(3) I've been talking about my RSS reader, my three-sided cards (see the media of my mastodon to see them at work) and other projects, I really need to get back into blogging. My plan is to set up Pelican, which is a Python-based static site generator. I've convinced myself that I can write Pelican plugins to make my blog exactly what I want, I really have to wire up a Jamstack setup for Pelican and just get it done.

I spent two days this week writing my first from-scratch Go program. All it does is periodically pull a JSON file from a defined set of web servers, parse that, and serve up a webpage with a nicely-formated table of the results - but it's exactly what I needed for that very specific purpose.

It's quite basic, but I'm quite pleased with how it turned out, as well as all the things I learned along the way.

It's August, which means fantasy football season is spinning up, so I dredged up my salary cap league program that created an ideal/efficient team using an genetic algorithm, updated it for my league's newest rules, and added in k-means clustering algorithm to generate tiers/buckets of similar players.

I’ve been building a searchable video library of BJJ instructionals and YouTube videos with transcripts (mostly done using whisper with some manual corrections) and an index of techniques tied to the curriculum at my gym. Unfortunately it is filled with copyrighted material so I can’t make it public, but it is going to be pretty useful for me as a lesson planning tool.

I've been wanting to start a side project, but I have absolutely zero ideas. I really like those online riddle games, so maybe something like that but as a mobile app? I did a side project in React Native a while ago and wouldn't mind brushing up there.