Why HA doesn't have a LTS branch is a mystery to me (other than the obvious maintenance overhead - but the community is huge and I'm sure they could find a maintainer).
When you consider the lifetime of homes, the hardware, etc it makes more sense for HA to have an LTS branch than almost any other piece of open source software.
I, for one, would really appreciate getting bug fixes, security updates, etc over a span of years for some installations. HA development pace is very impressive but the breaking changes can get old really fast, especially when you consider how much you can come to depend on the (ample) functionality.
When it comes to doing updates I typically allocate at least a few hours (just in case) to work through any breaking changes. I usually update ~monthly and like any rolling release strategy it helps minimize the number of breaking changes you experience at any one time but it's still a somewhat precarious situation.
Master needs to be the LTS branch, at least for all things mentioned. Once I have an automation it needs to work for as long as I live in the house. I have better things to do than to maintain the system. I probably won't upgrade the server at all for many years until I buy some new device that isn't supported by the old version, and I expect my house to still work after that upgrade, without having to spend days trying to remember how that old automation was programmed.
This here is exactly why the "smart home" is basically failing, because it needs to be set and forget and continue to work with ZERO maintenance beyond replacing dead components.
The fourth or fifth time you have to explain to your spouse that the lights in the living room can't be turned on because the app is updating or the server is down or whatever, you're going to really be feeling the desire to remove all the smart home stuff and go back to kerosene lanterns.
Which is one of the reason that I've slowly begun to phase Hue bulbs out - they work very well for what they do, but they can fail in annoying ways when the Internet is out, requiring you to get additional switches that can talk their protocol even when Siri isn't working, and at that point, why not just use a "smart switch" and dumb bulbs? At least that fails to just be a normal switch.
Agreed, though I have to say Phillips Hue lights have been well worth the money in this regard. I have some Lutron switches that snap over my standard flip-switches which prevents accidentally cutting power to the bulbs, and the switches talk directly to the lights (or perhaps through the hub, it's been a while since I've set it up) rather than through a server.
With HomeKit this all also works as long as my local network is up, with no dependency on Apple servers except for off-network access.
My ISP has had several outages, but I haven't run into the issue of "the lights don't work" at all yet. Only thing approaching this is I've automated them to turn on/off when I arrive at and leave home, so when I have people over and step out to run a quick errand all the lights turn off while they're at my place.
Hue works the best of all that I’ve tried for sure. But when we use Siri to control the living room lights, Internet access is needed. The Hue app still works if needed (most of the time) and the little no-battery physical switches keep working with Hue.
But it just doesn’t provide “value” compared to the cost, so I’m not going to rip it out but I won’t maintain it further or expand it.
The two need to work together. The switch on the wall needs to be smart as anyone in my house needs to be able to use it to control the bulbs, without having to open an app. Okay, if you can ensure a thief cannot control my lights while I'm my house I'm good with that, but I have kids, guests, and sometimes I don't have a device on me: it must work for all of them the first time and every time. A bulb cannot do that alone, it must have a smart switch (I've used the flip the switch off/on to control the bulb - it is not an acceptable work around).
The bulb however need the ability to charge color temperature and brightness depending on what I want, and a smart bulb seems best (though dumb dimming bulbs can to work well enough)
What I'd love to see (and nobody makes, but I could make one myself with a bunch of switches I guess) is to have various plain bulbs of different colors, and control them with the switch.
I wish the Hue bulbs worked with a actual smart switch (I have some of the snap on cover ones and they are "ok" but unsatisfying long-term.
There's commercial options that work like this (and indeed those installations can more or less be updated and upgraded with new stuff decades down the line, we've been there, it's been done, and continues to be done) but that's the rub: it's commercial, it costs money, and the hardware involved is much more expensive than cheap IoT gadgets out of shenzen. Wanting the lowestest cost option to also have the longestest life and highestest reliability is of course something we would all want, but generally not how this reality is structured.
When you consider the lifetime of homes, the hardware, etc it makes more sense for HA to have an LTS branch than almost any other piece of open source software.
I, for one, would really appreciate getting bug fixes, security updates, etc over a span of years for some installations. HA development pace is very impressive but the breaking changes can get old really fast, especially when you consider how much you can come to depend on the (ample) functionality.
When it comes to doing updates I typically allocate at least a few hours (just in case) to work through any breaking changes. I usually update ~monthly and like any rolling release strategy it helps minimize the number of breaking changes you experience at any one time but it's still a somewhat precarious situation.