Hi,
I’ve been using HA for several months now (since maybe January, or Feb of this year), eventually looking to use it to replace my other, more popular assistant. I have several integrations for my lights, TVs, doorbell, and more. I run this in a VM (VirtualBox) on Kubuntu 20.04, without the cloud integration.
I’ve not read release notes on a single update that I’ve done, and never once have I had an update that broke anything, and I update as soon as I see there is an update available to anything that doesn’t automatically update, but I don’t want to be bothered spending 10-20 mins reading through release notes every other week, before performing an update, when I have other stuff to do.
More than once in this thread I saw the words “intentionally breaking” and that doesn’t exactly sit right with me. It makes no sense to intentionally break a feature/integration, etc on an update, unless you’re updating that to fix what you break at the same time. (I could’ve misunderstood this)
I’m a decent IT guy for Windows, and a bit for Linux, but for this system, I’m just an average user who doesn’t know the advanced stuff, and would have to research to do anything other than use it, and run the very obvious manual updates.
I can get why some people would hate the idea of automatic updates, but even Linux has options to automatically apply updates, should the user want this. (Mentioned because I saw it mentioned somewhere here that Linux doesn’t do this, and I help manage a Linux server that is setup to do exactly that.)
I propose a compromise… Provide an optional integration that allows for automatic updates, should the user want them. If you put out an update that is an intentional break to an integration/feature they’re using, have the automatic update integration flag the update, and notify the user with an in-your-face warning that requires a manual update for safety (for those who do not read release notes). This way, non-breaking updates can be applied automatically, and breaking updates would still require a manual trigger, along with an obvious warning.
I get why devs would be inclined to refuse this idea, but being brutally honest, this is what the average consumer wants, when it comes to home automation/assistant software like this, and I think this might be a decent compromise for both sides of this.