I ran almost 24/7 for about 4 or 5 years in a commercial setting, and it's tinkering-friendly because the core feature is "modules" that are bundles of Python and HTML that you edit via a web IDE without needing to restart.
Plus, the device integrations abstraction is super easy and it's easy to build your own host that can use any device drivers you make.
We've also got modules for live sound mixing with JACK and CCTV cameras with ~1s latency to the UI.
I would say my home assistant (on a Pi4) is stable maybe 95% of the time, with most issues solved with a reboot. The biggest of those rare issues is that my zigbee dongle occasionally just refuses to work which isn't really an issue with HA its self.
I tried OpenHab in the past and was not a fan, but thats not to say its any worse.
But dashboard sucks beyond imagination. If you need any fancy UI - stay away. If you need pure automation - take a serious look at it.
I like it. Never tried honeassistant though, so can’t compare.
I would say: firm no.
See my other comment in the thread as to why.
Doing upgrades means things might break, but that is kind of understandable for such a extremely active project. ~10 new issues, ~25 merged PR's, ~4 new PR's per day for the last month.
I wish there was some way to better decouple the new things from the core though, hopefully they will figure something out so we can have stability without opting out of new features completely.
The comments indicate it is a much older java project, so perhaps it doesn't do anything better?
The rate at which home assistant moves is probably its biggest strength and also its biggest weakness. Every time I upgrade something breaks and requires debugging. It's one of those things I wish I could setup and never touch again, but alas, I like to tinker...