Unique ID

[ronaldt80]

George, any particular reason why you removed the unique id from this component? If not, mind if I launch a PR incorporating this once more?

[GeorgeSG]

Yeah, I removed it because it was not working consistently. It depended on acquiring the Mac address of the device via the API. I don't know if it will even work over COAP.

People used to have issues adding multiple devices, which was also due to the unique_id implementation. This may have been a bug, but it was hard for me to debug it (no 2nd device), and I know the current approach w/o unique_id works.

If you want to change the entity id, you can set the name property.

If you can implement it in a stable way and verify that it works, I'm not against reintroducing unique_id.

[ronaldt80]

I raised the issue initially on the original implementation. Your former development branch however solved my issues and works with 2 machines. I have had it run consistently for almost a year. Does indeed not work with coap though but betaboon has found a different solution, will try to reverse engineer this. Otherwise perhaps use a random number generator instead? Or link it to the last digits of the IP address?

…

On Sun, 11 Apr 2021, 11:55 Georgi Gardev, ***@***.***> wrote: Yeah, I removed it because it was not working consistently. It depended on acquiring the Mac address of the device via the API. I don't know if it will even work over COAP. People used to have issues adding multiple devices, which was also due to the unique_id implementation. This may have been a bug, but it was hard for me to debug it (no 2nd device), and I know the current approach work. If you want to change the entity id, you can set the name property. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#5 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMXEM3A3F6WSTKVWDQOGHMDTIFW2XANCNFSM42XNRMHQ> .

[GeorgeSG]

I'm not sure about using random numbers. I think the guidelines are to use something tied to the device, and if the device has an IP, use the mac address.

There's a good discussion here.

Also, I see that betaboon's component uses their own custom python package. Please try implementing this via py-air-control, I'd like to stick to that if possible.

[GeorgeSG]

Also, I'm planning to convert this to config flow to be future proof.

[ronaldt80]

Also, I see that betaboon's component uses their own custom python package. Please try implementing this via py-air-control, I'd like to stick to that if possible.

I tried the py-aircrtl implementation of COAP; it is flawed and does not work well with the newer machines. The purifier crashes as a consequence of the continuous polling by HA.

I made it working based on your component, using the COAP client instead of HTTP; behavior was erratic - device was turning unavailable after 20 seconds and reappearing after a minute or so. The pyairctrl output (CLI) is not much better.

There are quite some discussions on the HA community forum on this.

[d4g]

Is there any update on this?

[ronaldt80]

Is there any update on this?

you can check my fork - I implemented the unique id based on Mac Address. Works fine for me. It's a requirement to have a unique ID to expose the purifier to voice assistants.

BTW - don't use my complete fork unless you have the AC3259; I hard coded the sleep mode specifically for the API of this model.