Hector Martin

I was going to write a @home_assistant integration for my aircon. Then I found that project, and figured maybe I didn't have to. Still going to do it.

I just... don't understand how multiple people contributing to a project end up (manually!) capturing thousands of IR codes from remotes to make this work and don't stop and think that there might be a better way.

I'm interested ... is it possible in simple words to explain how it could be done? Isn't it necessary to try the keys one by one anyway?

Aircon remotes don't send keycodes, they have a screen and keep track of the current settings on the remote, then transfer the entire state to the aircon. The right way to do this is to reverse engineer the packet format and figure out where mode, temp, fan speed, etc go.

These people are literally spending hours with each remote trying every possible speed setting times every possible temperature times every possible mode (and missing out on any additional features because some have so many combinations it would take weeks to capture them all).

For reference, here is a quick emulator I wrote for my old aircon remote: https://mrcn.st/p/jUaSBl4E  It really isn't hard to work out the fields. You change the temp, see what changes; change the mode, see what changes; etc. Often you can get more modes than the remote supports.

That's nice, especially if you can get feedback via that. With IR I'm flying blind...