Ask HN: Share your side project

Mine aren’t even half baked, more like still raw. Maybe a hundred lines of code written between them. I love to talking about them though. The few “active” ones though are:

1. A user mode ISDN stack written primarily in Erlang, with bits of C and Java for certain components. This one is pretty ambitious but it will end up supporting both typical LAPD/Q.931 connections as well as Frame Relay and X.25 on both BRI and PRI interfaces. I’ve sunk more money that I should have on hardware for this project and still need more. At the interfaces level, though the stack will support DAHDI devices (if I can ever get it to work properly), the osmo-e1d interfaces, which is a neat price of hardware I found about about through HN[1], and whatever normal ISDN terminal adapters I can reverse engineer control protocols for (I imagine most are at least somewhat AT compatible). Biggest barrier for this project is just my laziness and unwillingness to write some of the more boilerplate stuff as well as handling the U-interfaces for BRI since it’s just not possible to get an actual ISDN BRI line anymore. I’ve wondered about hacking around it with NT1’s or U-interface expansions for Cisco routers, but I have my doubts it would work. I could also use some CO simulators but they’re rare and other line emulators are rather expensive. Its hard to get help when I get stuck because it’s just so niche. Can’t really reach out to most retro resources and most telecom stuff around those seem mostly analog/PSTN type stuff and more modern networking/telecom communities will likely just play 20 questions trying to figure out why I’m using an obsolete technology. Like I said it’s an ambitious project, and I’ve planned out extremely good design docs, reference applications including a H.320 video soft phone and a couple of APIs to interface with it (CAPI and JTAPI). I’ve probably mentioned this project so many times that people don’t actually believe I’m working on it though.

2. A custom DOCSIS modem, built on a limesdr. Unfortunately I only have a very old DOCSIS CMTS and cannot run this on a real line legally so it’ll be based on whatever DOCSIS version is supported by my CMTS. Haven’t done much work on this because there’s just so much setup to do. I need to reset the CMTS, set up the sdr, brush up on calculus, keep the manual for the CMTS and DOCSIS specs open to refer back to, etc. I also plan to make Linux and Windows (NDIS) drivers so you can connect the sdr directly to the coax and get a network interface.

I’ve been building myself a custom running dashboard to track my stats over time. It’s a lot more fun than paying someone else and means I can build up whatever visualisations I like. The current version is at https://running.callumm.dev

Overall the project consists of:

- An iOS shortcut that converts the day’s data to a CSV using HealthExport [1] and submits it to my API

- A tiny Phoenix API [2] that uploads the data to graphjson.com which I use for the visualisations and to avoid maintaining a database of my own for this

- A dashboard written in NextJS that shows my visualisations, hosted at that running.callumm.dev URL

I built it on a whim over the last month when I decided to start trying to run every day, so it does fun things like compare to last month. I’m sure I’ll keep tweaking things over time but this stack feels pretty good to hack on so far!

My favourite visualisation is one that shows the proportion of time spent in each heart rate zone per run over time. It’s something I’ve been working to get under control (I used to run at a consistently very high heart rate) and it’s awesome to be able to visualise that progress.

[1] https://healthexport.app/ - not mine or affiliated, but it’s a couple of dollars one time buy and makes my iOS shortcut work!

[2] GitHub repo for the data import API: https://github.com/mcintyre94/callum_runs

[3] GitHub repo for the dashboard: https://github.com/mcintyre94/callum_runs_dashboard

I have been doing software development for over 15 years, and I thought that I could apply some of my skills in another dev domain for fun. This time I wanted to try something different, and I started working on a tiny 1-bit retro game. I heard many horror stories that gamedev is hard, and I wanted to get the heartfelt first-hand experience. While it is certainly true that prior experience helps, but gamedev requires a multidisciplinary approach. One has to compose music, create sound effects, draw tiles and sprites, think carefully about game mechanics and many many other things.

On top of that, I am a solo-developer, and I have to prioritize ruthlessly, i.e. whether a new idea adds something meaningful to a game, or it is nice to have but will drain my energy, and I will abandon a game quickly. I am lucky to meet great people in gamedev groups. Also, I find gamedev community very friendly and supportive. I am still miles away from the first playable build, but it is a rewarding experience. If anyone is curious, here is what it looks like: https://twitter.com/oneearedrabbit/status/1420575365131653126

I have been working for a few years now on Exomind[1], a personal knowledge management tool that takes the form of a unified inbox in which you can have your emails, tasks, notes and bookmarks organized into collections. I have iOS and web/electron clients at the moment and a simple browser extension for bookmarking.

Its backend (Exocore[2]) is built on top of a personal / private blockchain and is made from the ground up to be hosted in a semi-decentralized fashion on your own personal devices (your computer, raspberry pi, a cloud instance, etc.). It is written in Rust and has iOS, C and Web (WASM) clients. It's extensible via WebAssembly written applications.

It has very rough edges, but I'm using it daily to organize my life. It has also been my learning playground to improve my Rust skills over the last two years (it was on another tech stack before).

[1]: https://github.com/appaquet/exomind [2]: https://github.com/appaquet/exocore

https://github.com/abhijat/matchboxdb

I'm trying to build a database system from scratch (without using third party libraries) in modern C++ (I do use googletest for unit tests).

I've mostly been stumbling around but having no prior experience writing anything like this, I think I am at a place where things might start coming together soon.

Initially I had very different goals but now I think I will focus on things like indices, full fledged SQL support and more data types later, once I have the basic model working.

I did get to learn a lot while working on it, and hopefully will get to learn a lot more as things come up.

Some of the resources which were very helpful were the CMU database course on youtube and the book called "database internals" by Alex Petrov (I just read parts of it as required). Also the book Writing an interpreter in Go by Thorsten Ball was very useful when writing the sql parser.