What programming/technical projects have you been working on?

Had to stop studying Python for a while to make a website for my mother’s company. Unpaid work but putting me in the world has its perks.

A good opportunity to really learn Flexbox and CSS Grid. Bootstrap is too messy for my taste.

I pilfered a javascript version of pacman and have been modifying it with my son to teach him rudimentary programming concepts.

I created a couple of custom maps that can be edited via Atom, along with altering the number of lives so that the pacman is almost immortal.

We also changed the pacman, sometimes pacman is pacman, sometimes the pacman is my head, and sometimes the pacman is his head. The death sequence animation was the most fun, and I still bust up every time I see the crazy faces my son made to indicate the pacman was dying.

I just ordered a PineTime dev kit to learn how to work with embedded systems. This should be fun!

I'm working on a mac menu bar app (full stack python made possible by the rumps library) to add tracks playing on WFMU to Spotify. I'm also working on functionality to subvert the recommendation algorithm with further reaching but within bounds recommendations. I suppose it's a harder problem to solve computationally but I'm not strictly interested in music like the music I'm currently listening to. It's much more nuanced than this in reality. Probably not the right thread for this but does anyone else feel similar?

I got distracted from my business this week to focus on building a Paw Dynamic Value extension to retrieve a specific value from a database. The tokens I use to verify a user and organization in our applications are not accessible from the client (rightfully so), which means Paw has no way to know what the value of the verification token is when submitting a Verify User/Organization HTTP request.

Unfortunately for me, I've encountered several limitations in the Paw extension interface that prevent me from implementing this as nicely as I'd like. Paw utilizes Safari's JavascriptCore engine instead of running something like a containerized Node.js environment, which means I have no access to libs like fs and net, which tools like node-postgres need to work—after all, it's not usual to try and interact with a database in a browser environment!

The solution to this appears to be extremely hacky: run a DB access server on some local address, and have my Paw extension make HTTP requests to that to retrieve the value I need. Yuck.

I have posted on tildes before about my interest in the history of public access unix systems and in the future of the non-commercial internet.  My project recently has been to set up and run a free Linux server for people interested in learning about and practicing command line skills and shell programming in a collaborative social environment.  It's an opportunity for me to both support the non-commercial internet and to level up my sysadmin skills.

Although users can publish web and gopher content to the web, most activity is internally focused.  And most tools for user interactions are built by the users themselves.  There is a user-designed shell blog (called "shlog"), a command line chat program, a git-based bulletin board, and a few other novelties.  I am currently working with a couple other users on a distributed link aggregator as well.  A cool part of most of these tools is that they are built on a decentralized architecture -- user content is stored in each user's home directory and aggregated into a view by client tools.  Because most of these tools are internally focused, it gives people a place to play with computer and internet technology without broadcasting everything to the entire world.

The technical side of this project for me, again, has largely been about leveling up on sysadmin skills.  I started the system on a Debian VPS and have recently moved to Arch.  I'm now running an nginx web server, a gophernicus gopher server, postfix for email, and various other security and other configurations and services.  Although I'm philosophically not a fan of systemd, there is no denying its importance, so this has also been an opportunity for me to learn more about it.

"Are you having trouble with anything?"  Not trouble exactly, but one of the challenges (in the good sense of the term) is resource minimalism.  The VPS is tiny and will stay that way intentionally, so a theme of the system has been about doing as much as possible with as few resources as possible.  The fun part of this is, as many of you may know, is that text-based computing is lightweight, so the system isn't nearly approaching any limits yet.

This is intended to be a long-term project so maybe I'll post again if anything notable happens.  (And if anyone is interested in joining, just PM me. The system is new and has a small user base, so there is a lot of room for more members.)

Re: Unsubscriber, it worked, I think? The day after I still got a lot of email, and I went through manually Unsubscribed, but then I still got email and apparently it may take up to 10 days to properly Unsubscribe for reasons. So I dunno, it did get me less email, but that might have been because I did it manually.

Also of note, there's an at work Hackathon for trying to make a Trello drone fly around with a voice assistant, using the provided web tools or C API. I'm off Friday, so it's another 5 hours I would be spending at work, but hopefully I can spin it into a raise. I mentor through FIRST Robotics, so I think I can make something happen.

I put together some Python code the other night because I'm looking to purchase and AMD 3950X and did not find a Now In Stock monitor for Microcenter. This gave me an excuse to learn a bit about Beautifulsoup4 and also the Twilio Python API so that my script can send me a text when the CPU is in stock. I expanded the code to be able to look through all Microcenter locations for any number of products...though I don't know if all product pages will work. It's basic but a fun way to try some new libraries. It's command line only and now real use case outside of meeting my own personal need to poll for product availability.

I am attempting to do a 64-bit port of cyclone, which was doing safety before it was cool.

I've managed to build a copy of the compiler under 32-bit, which was no small feat. Unfortunately, I'm having difficulty bootstrapping, and have already had to make some manual modifications to the autogenerated c code. The main problem is, the automatic c binding generator seems to be unable to parse stdio.h (and a couple of others, but that's the main one). There are stubs for the minimal interface that the compiler needs, but the parser generator wants more.

I’ve added four 4k cameras to my off road robot and I’m learning to write a gstreamer plugin to process the data. I’m going to collect a bunch of data, hand label the images, and run pixel wise image segmentation on the images. That will let me detect and follow trails. I will also use structure from motion techniques to estimate the 3D depth of the scene. The real trick will be getting a good frame rate! Some pictures of the system and output are on my website here:
https://reboot.love/t/new-cameras-on-rover/

I'm working on beefing up my home lab setup so I can maybe make some vulnerable VMs to provide back to the community, and on the side of that side project, starting to write an updated version of cewl to scrape websites and generate possible account passwords based off of the content. I have a really shitty version of the manipulation engine sketched up in python3 so far and it's been sitting on the back burner :(

I'm currently writing a library to decode, encode, and verify cryptocurrency wallet addresses.

I was really impressed by Bitcoin's Bech32 encoding scheme. It would be cool if this was adopted outside of the cryptocurrency space too, I think it would be great for encoding something like a Tor onion address.

I've been plodding away at getting elm 0.19.1 running on my Raspberry Pi. Unfortunately there's no precompiled binary for ARM. This has led me to dive into installing haskell and trying to figure out how to build cabal-install from source. There's been lots of small issues, e.g. version mismatches of base libraries, leading me to install other versions of ghc than 8.2.2 which was indicated in a blog post to be needed when installing elm 0.19.0.

So after getting to the point of having ghc 8.4.4 and a working cabal-install on the Raspberry Pi I try try to build elm... only to discover that elm requires the time-1.9 haskell library which conflicts with the requirements of the ghc I have installed! Aaaah software dependencies, it is almost as easy as naming and cache invalidation ;)

Armed with a little bit better knowledge of ghc/cabal, I now have to rebuild cabal-install, but to do that I need a newer version of llvm! One step forward two steps back...

For some reason, I've become obsessed with Penrose tiling. So I've implemented a couple different apps to generate tilings. The first does a recursive search, laying down a tile, then laying down another following the edge-matching rules, etc. until it gets to a point where it can't lay one down. Then it backtracks and tries again.

Next I implemented the much easier method of inflation/deflation where you start with a tile, generate a tile that contains that tile, etc. This is fast and easy for a small area but it makes it difficult to pan around and expand the area covered by tiles.

Finally, I implemented the pentagrid method which was difficult to find information on. All the info is either mathematical papers, or descriptions by people who leave out important little bits, or describe important steps sparsely. I eventually got it working, though, and it's pretty cool! I still need to label the edges after generating the rhombs so I can convert them into darts and kites or pentagrams and pentagons. But I'm having a lot of fun with it.

Just been working in RStudio for some data projects.