yeah i think chicken/ egg issue, those with mqtt setup or a new users.
i am not sure how the new integration is playing out, but you may find it actullay preferable to move over to the integration method for a cpouple of devices to test or any new ones you add
I think if the end device supports mdns its would be the default option to set zeroconf,
not all devices support mdns so are unable to be discovered or are external services thats why they have the orange +
yaah but I prefer messaging method (mqtt) over polling. it’s better from various standpoints: architectural, maintenance, flexibility). IMO a lot of users have similar attitude.
might be true but then I feel it counterintuitive.
I always thoght it’s just enabling another module/feature explicitely.
i guess, but also depends on the device, i have seen some expose far more functionality via their api than via mqtt
Another way would be possibly to split the integrations page, those which are discovered but not setup (problem would be notifications) and another page with those that have been explicitly setup via the + or discovered and have been configured
ShellyForHass is a custom component, not official one.
So open an issue and propose better solution. At the moment discovery/zeroconf works this way for every integrations using config flow.
No. I’m going to script it as I do with my all shelies configuration. It must work even for future devices, right?
Wonder why it doesn’t seem to you to be a lot work if you multiple number of people using mqtt integration with with all this unnesesery clicking. Some guys have more than 100 devices (so it’s more than 200 clicks). Are you serious?
And then more after some devices appear from deep sleep and again with every new device added to the network (and again multiplied by number of users who use mqtt to connect shelly devices).
You obviously have problems admiting that forcing anyone to clicking hundreds of times in integration never chosed to be used is just plain wrong. As I wrote in other thread you guys rather advocate for issues to death rather than say sorry and re-thing an approach to get rid of such glitches.
Yes I’m going to file issue in GitGub, but IMO the developer itself should be the first who should do that, cause he knnows the impact the best. But again “it’s not integration issue” so he feels no responsible for what happening. Another attempt to move responsibility to the user who obviously should stop using mqtt or disable some features of OWN system (mDNS) to avoid issues in HA… or click to infinity on integration I don’t use.
Seriously: Do you really think this behaviour is correct?
I think you are making a mountain out of a mole hill.
I was just trying to get it sorted for you in 30 seconds. But hey, good luck with the issue and custom script development (and debugging).
1 Like
I too constantly get Zeroconf identifications listed in my integrations page - getting rid of them is a daily irritation. As Zeroconf is part of the default config it is not something I am happy to remove - if the developer intends this behaviour then I am a bit surprised.
The idea of zeroconf is to make it easier for new users as part of onboarding or when someone add a new device is found. If it was is disabled in the integration they would never be found.
so for new user or you want a new device to be found easily - zeroconf in a good thing.
Those with existing devices setup and not using the integration - zeroconf can be a pain as it will find them
The default for integrations using config flow is to have zeroconf enabled for it
The choice is you disable zeroconf in HA by pulling apart default_config and adding each item listed in default config separately but leaving out zeroconf, or disableit in the end device if possible or hit ignore in the integration
Again i cant see how you can change this behaviour
The Zeroconf thing seems to make a record of the devices I mark ignore - but doesn’t seem to use that record to stop asking again in the future: would that not be the simple enhancement needed to solve the problem?
If I was HA author I would really consider getting rid of zeroconf (at least for such integrations). From my experience systems which try to guess what do user want ussually cause more pia than expected. At the end every bit experienced user disables it.
If it must stay as is I can see following suggestions (feature requests)
- easier way to disable zeroconf for selected integrations. current way requires getting rid od default config and defining it manually (it’s not hard to imagine potential impact in future, besides it’s unnecessary effort). I think I saw it in WTH
- option to disable integration at all untill I decide otherwise.
- option to ignore all devices discovered by particular integration by single click (something similar to what HACS offer but limited to single integration to make it even better)
Thank you. I’m smart enough to know that I can hide devices by Ignore button.
The reason of this discussion is to
- point to weaknesses of current state
- discuss possible options to workout in more comfortable and/or persistent way
- come with potential ideas of changes which might then be a source for feature requests
if you hit ignore, its not supposed to ask again, if it is i would log a github issue
Okeydoke.
I don’t suppose you can point me to the right place?
Yes because:
- Your problem is not related to Shelly integration, but HA architecture and I can’t do anything in integration code to fix it
- You didn’t open the issue in the appropriate repository, but you blame everyone that something works in a way that doesn’t suit you
You probably forget that most of the people who work on it do it for free, in their spare time, for fun. Instead of starting unnecessary discussions and accusing others of not being willing to help, try to help yourself. And yes, opening an issue with a detailed description of the problem is also help.
1 Like
Yes you can. You can raise the issue yet before delivering your work. It’s because you are possible the first person who know impact of your component on the other parts of the system.
I did open no issue yet. I started the discussion to get to know what is the reason of this behaviour, what parts of system are responsible for it and learn about potential solutions. Are you getting this issue too personal? Correct. You should.
Nobody is forgetting, neither me. I simply refuse to accept it to be an excuse everytime something goes wrong.