Hacker News Reader

Comment by snapetom | Hacker News Reader

God, don't use Homeassistant if you actually value your time. I've ranted and raved here on HN about it before. It's a hobby project where the devs only care about using the latest and greatest methods and tools for the sake of padding their resumes. They will often re-write things for no reason, release things into the UI as placeholders, and don't care how many end user installs break. You will spend massive amounts of time keeping up. Case in point - they removed Python 3.7 support in late 2020. 3.7 is not EoL until mid 2023, and was the default version in Buster, the Raspbian release at the time. We all had "fun" compiling Python from source.

Just get a Hubitat and be done with it.

View on HN

They're also hostile to distributions packaging them: https://news.ycombinator.com/item?id=27505277

Holy crap! Home Assistant is such an interesting project, and I was seriously considering it for the home I'm relocating in probably this summer, but that thread almost makes me want to avoid it straight.

Thank you, I completely missed that. I am not surprised at that attitude given how badly run of a project HA is.

I agree about all HA critics BUT what can you suggest? OpenHub is even worse, behind HA I can't find anything with a ready-made WebUI and ModBUS/MQTT pre-coocked support...

I use it since a bit more than an Year, pip-installed and kept updated on my homeserver, for p.v. (Victron + Fronius + Giavazzi + ... all ModBUS/tcp or MQTT), it's config is obscene YAML, but with org-mode (org-babel + noweb) is manageable, logging is obscene but work OOB enough, automation is obscene but work enough, if logic is not much complex it's doable, the WebUI is nice...

I know I can write something myself, Mosquitto is FLOSS, ModBUS FLOSS libs are there for various languages, but that's not so nice for a simple domestic use case like just seeing how a small p.v. plant works, control the VMC/heat pump and hot water...

Just to play with sensors and crappy wireless switches is even LESS motivating...

IME IoT is crap in general, mostly because too many have no use-case for it so it's sold just to play, not to seriously use and profit from it, also because it's mostly designed by electronic engineers not IT techies, and we all know how ee are bad software side, many other reasons but well... HA so far for me was/is the leas worse pre-made/almost ready to deploy stuff I found. Surely I do not care at all at their suggested deploy with docker, HA cloud etc, just plain pip, but nothing have stopped me so far...

Sounds like Home Assistant is aggressive about keeping up to date. That’s a good thing!

Breaking user installs is not aggressively keeping up to date. Have a blank map in the UI that didn’t work for years is not aggressively keeping up to date. Rearchitecting major components three times in five years is not aggressively keeping up to date. Creating, recommending, then removing an install method all within about 3 years is not aggressively keeping up to date. Having four different ways to configure things is not aggressively keeping up to date.

It’s having no discipline nor foresight and instead throwing shit against the wall to see what sticks.

I bought a "conbee" (basically a USB Zigbee dongle) and after setting up their DeConz software and detecting all my devices (IKEA, Philips, Lidl) I never bothered with Home assistant. Or should I say, I started to, but gave up because it was a pain in the neck and anything it offered beyond what DeConz did, wasn't worth the hassle. I always meant to go back and have another go, but over a year later I simply don't have the motivation.

Were you doing a supervisor install or something? I didn't even notice the python upgrade. I do occasionally get some breakage when I upgrade versions, but it hasn't been very often that I would even consider using something else.

Venv on core. Yes it’s the “expert” install. Yes Docker would be easier. However you can’t guarantee Docker will be supported in the future. They created then dropped Hassbian like a hot potato. In fact there are already signs Docker is falling out of grace in favor of HassOS.

That is specifically called out by the ha team as an unsupported and not recommended type of install. So I think it's a bit of bad form to be complaining that your specific unsupported install isn't well supported (and on a free software, even worse).

It's everywhere in their documentation.

https://www.home-assistant.io/installation/ https://www.home-assistant.io/installation/raspberrypi#insta...

It's not "unsupported" if they tell you how to do it. In other forums, devs have also promised that core will "always" be an option.

It was also the only installation method just a few years ago. Then they experimented with Hassbian/HassOS/Docker, etc. Shit, I remember they had a VDI option less than two years ago. That was FIVE simultaneous installation methods they supported. They shouldn't have let it get that bad to begin with, and pruning just goes to my original point that they keep changing things because they never have any discipline in the project.

I just turn off automatic updates and look at change logs to see if it's worth it to upgrade. I also primarily use HA as dashboard and sensor fusion, it only runs a couple automations directly.

Is Hubitat actually better?

Hubitat’s UI looks like it came from the early 2000s. There’s a small learning curve to get used to its philosophy, but it’s pretty consistent. Once you get it, it’s stupidly easy.

I mean for your case, as a dashboard, HA might be better because the Hubitat UI is pretty ugly.

> Just get a Hubitat and be done with it.

Am I right in concluding this appears to be another closed-source brick which is likely to implode in a few years, as they always do?

I have used Hubitat to control lighting, shades, and locks in my home since 2018. There is no reliance on outside services or an internet connection. If the company ceases to exist my device will continue to work at least as good as it currently works today. However, there are many third party drivers written so it is conceivable development could even continue.

You CAN interface with other outside services if that's your jam, but at that point relying on Hubitat continuing to exist is no different than relying on those other services to exist.

No because there’s no reliance on a cloud service. There’s cloud features, but there’s alternatives and workarounds that they don’t (can’t, really) block and even suggest if you don’t want to pay.

Reply on news.ycombinator.com