Very bad idea which unfortunately is based on incomplete data. In ADR-0019: “All of these integrations have a low usage count (source Home Assistant Analytics)”
231 out of 122 thousand is irrelevant in terms of number of users.
If you feel that ESPHome and similar aren’t options then the solution, of course, is for you and others that want to keep them to step up and maintain them as custom components.
So 80% of integrations are irrelevant in terms of number of users?
Why accept others?
HA must be limited to 50 integrations and especially not put on the website “Works with over 1000 devices” if more than 800 integrations are not relevant in terms of number of users and can disappear at any time
or maybe because no maintenance has been necessary ?
Maybe that is part of the reason that Raspberry Pi Foundation have not yet released 64-bit RasPi OS … and surely RPF need to get GPIO integration working in 64-bit RasPi OS before anyone on this end can start to test or maintain it ?
Torok42, this seems to me very much more part of a trend to position Home Assistant as a service running on a PC in a back cupboard, with all sensors connected via network.
Raspberry Pi is still a cheap way for people to start with a server, but we have to stop thinking of it for its I/O.
A separate physical RasPi may be necessary to act as a remote I/O controller device, separate from Home Assistant. I note that there is a remote_rpi_gpio integration … but I don’t think it really does this.
Also it is extremely easy to copy an existing component folder and move it to custom_components … (the last time
I did it anyway it was)
If are are really set on keeping it going I would suggest you make a GitHub repo for the “custom_component” which can literally just be a copy of the existing one… and maybe your post will catch the attention of someone with the capability and desire to keep it functioning… but in the meantime that should still function on your system even beyond it’s deprecation … (other than the aforementioned 64bit issue )