● /sudo-dev/timeline
The trail
so far.
This is not a résumé and it is not a claim that every stop was finished. It is the sequence of problems, tools, and rabbit holes that shaped the working notebook.
// the first pull
One garage door.
A bad myQ workflow turned into Homebridge because Siri should have been able to do one simple thing. It worked quickly. That was the trap.
// the rabbit hole
Home Assistant became the center.
The question stopped being “can I automate this?” and became “what does the whole system need in order to stay understandable?”
// local infrastructure
Linux hosts, storage, and services.
Self-hosted services, Samba, media, backups, and the ordinary maintenance that turns a spare machine into infrastructure.
// the network underneath
OpenWrt and home networking.
Routers, DNS, remote access, segmentation, and the connective tissue that makes every higher layer either feel reliable or mysteriously broken.
// more radios
Zigbee joined the stack.
Zigbee2MQTT made the hardware visible and gave the lab another protocol to understand instead of treating every device as magic.
// another network layer
Thread and Matter arrived.
Separate credentials, border routers, and ecosystems made the home stack a real systems lab rather than a collection of smart devices.
// the board was not the only thing assembled
An Ethernet Thread build.
An ESP32-S3/H2 experiment became a study in board selection, SDK versions, radio compatibility, and the very specific ways an embedded build can refuse to cooperate.
// the ecosystem problem
Apple credentials and border-router behavior.
HomeKit, Apple Thread credentials, and OpenThread did not collapse into one neat diagram. The failures were useful because they exposed which assumptions were wrong.
// the web work
Small sites and deployment trails.
Web projects, DNS changes, Cloudflare, documentation, and the unglamorous work of making a site remain understandable after it leaves the laptop.
// a different kind of native
SuperSpock.
The browser client worked, so I started a native macOS client for the GL.iNet Comet RM1. DNS and TLS are explained; the Pion/WebRTC path now negotiates and receives video.
// the reference implementation
The RM1 frontend clarified the protocol.
Reading the device’s shipped frontend turned a vendor mystery into a concrete Pion/WebRTC investigation. It gave the project a better boundary.
// current investigation
Keep following the evidence.
The subject changes: Linux, networking, automation, web, hardware. The method stays the same: build something real, break it honestly, and leave a trail someone else can follow.