-
-
Notifications
You must be signed in to change notification settings - Fork 97
[feat] Use trigger instead of sensor in home assistant discovery for scenes #738
Comments
Why?
Aren't the automatically created?
This is because you have not set retain flag I think |
This is more theoretical than a real concern on my side. Since the scenes are forwarded to HA by setting a sensor to on then off, I assume that when turned on there is a delay before setting it back to off during which, if the device trigger the scene again, the HA's sensor won't toggle on/off or will do it after that delay.
Yes there is an event which is fired each time a device state changed, but my comment was more about the distinction I make below.
Indeed, I forgot I let the retain flag unset in the settings page. I'm no longer a big fan of the retain, usually causes more issues than benefits (stale values that has to be manually cleaned-up). It might do the trick, even though I'm not sure for the scenes that I never trigger. Just to clarify, I've created this issue as I was trying to re-arrange my UI using the "auto-entity" filtered by device. I realized that I had those sensors that shows up but was absolutely useless in the UI (and just filtered them out still using the "auto-entity" card). It make me realized that the distinction between event (a.k.a. intent / trigger / command / or even scene in z-wave terminology) vs. state was weak when backed by a device in HA, as a device is state-full. Turning a sensor on and off is actually a patch IMHO, even if it's a common pattern that I've fight more than once in my professional life. So I looked in HA docs and found that the MQTT integration does offer a solution for that distinction which is well documented:
I won't argue that it's the ultimate solution that must be implemented right now, it's more a suggestion as this feature in HA has been developed specifically for those cases. And actually I agree that currently a sensor is really easy to use in the automation GUI (even if I don't use it) which might not be the case of events (I'm planning to use the "automation_trigger" for my own projects, and see if I've to open an issue / contribute to HA to ease that use). |
Ok thanks for the clarification @dr1rrb :) I will implement this, just one question, with scene you mean the |
Thanks @robertsLando 👍 And yep it looks like to be that: |
@dr1rrb What type should I use in the discovery payload? What value is published in the scene topic when it is activated? https://www.home-assistant.io/integrations/device_trigger.mqtt/#type Also could you try to manually edit a scene discovery configuration in order to work as a device trigger? I don't own devices to test this |
Hey @robertsLando , I finally took time to validate that. So for the discovery I used {
"type": "trigger",
"object_id": "scene_down",
"discovery_payload": {
"automation_type": "trigger",
"type": "button_double_press",
"subtype": "turn_off",
"topic": "zwave2mqtt/inovelli_lzw31sn_001/91/1/1",
"device": {
"identifiers": [
"zwave2mqtt_0xc35fa37d_node3"
],
"manufacturer": "Inovelli",
"model": "LZW31-SN Dimmer Red Series (0x0001)",
"name": "inovelli_lzw31sn_001",
"sw_version": "1.35"
}
},
"discoveryTopic": "device_automation/inovelli_lzw31sn_001/scene_down/config",
"values": [
"91-1-1"
],
"persistent": true,
"ignoreDiscovery": false,
"id": "sensor_scene_state"
} Note: No Then in the HA's Automation UI it adds a well named trigger: About the payload to publish, actually whatever you want. It's seems that publishing a value on the So it means that unlike other properties, we have to discover multiple HA entities for a single "ZWave class ID" (not sure if it's the right name, in means the "91-1-1" in my case). One for each possible value of a scene. Apparently we already have the info in the device config file: <Value type="list" genre="user" instance="1" index="1" label="Scene 1" units="" read_only="true" write_only="false" verify_changes="false" poll_intensity="0" min="0" max="0" vindex="0" size="0">
<Help></Help>
<Item label="Inactive" value="0" />
<Item label="Pressed 1 Time" value="1" />
<Item label="Key Released" value="2" />
<Item label="Key Held down" value="3" />
<Item label="Pressed 2 Times" value="4" />
<Item label="Pressed 3 Times" value="5" />
<Item label="Pressed 4 Times" value="6" />
<Item label="Pressed 5 Times" value="7" />
</Value> I don't know if I can already add some entities for each value by manually editing in the settings.json ? But even then, I think that ZWave2MQTT only converts each "ZWave class ID" into a given MQTT topic, I won't be able to request it to publish in some extra topic. |
BTW I just realized that if for the
So I tried to alter my discovery config to use the config's "discovery_payload": {
"automation_type": "trigger",
"type": "Pressed 2 Times",
"subtype": "Scene 1", That looks really great IMHO |
@dr1rrb Are you able to submit a PR for this? :) |
@robertsLando I would love to, but I already have a lot of pending projects (the life story of all of us ;)). So it probably won't be before the end of this year. Meanwhile I can do real world tests if someone wants to do this. |
@dr1rrb I keep it in my todos for now but I'm also working on z2m successor (https://github.com/zwave-js/zwavejs2mqtt) and it will take me some time :) |
Is your feature request related to a problem? Please describe.
Currently Z-wave scenes are pushed as "sensor" in Home Assistant, using a hard coded delay to reset the state when triggered.
This means that
Describe the solution you'd like
We should instead use the "trigger" component (a.k.a. "device_automation").
Describe alternatives you've considered
This could be an option in setup to avoid breaking change.
We could also have both enable as a migration scenario.
Additional context
Add any other context or screenshots about the feature request here.
The text was updated successfully, but these errors were encountered: