4 ms·
Having a mechanical fallback is a really good idea. I always get a little worried when I see people patching together mission critical systems on a raspberry pi
by kayodelycaon 4y ago
Having a mechanical fallback is a really good idea. I always get a little worried when I see people patching together mission critical systems on a raspberry pi with little thought to failure modes. Learned this is the hard way.
Dealing ~99% reliable smart bulbs and guests not understanding the fancy buttons is death by a thousand cuts. Even a 0.1% failure to respond will add up fast if you have multiple bulbs per fixture.
Eventually, I switched to using Lutron Caseta products as the base layer. Now my light switches will still act like light switches if the hub is down and all the remotes are dead. And the remotes don't need the hub online.
- adql 4y agoYeah there is very little thought in resilience in most such systems. Ideally I'd see some generic rule-based system (maybe just WASM VM fed with some code and triggers?) that does just that - feeds code some data and pushes output to controllers, maybe with standarized methods to ask for historical/other sensor data, then have that just be a blob you can deploy on few devices, share a password and it would pick a master and keep rest in standby. Then just have HA-like Web UI for configuration so the big fat of UI can run really wherever on anything, but the core that is needed for everything to actually work can sit on some cheap device in the closet.