Live data from Hacker News

I built an offline smart home

androidauthority.com

151–160 of 192 posts

Re: I built an offline smart home

#151
post #120

Earlier quoted context omitted.

Can you explain the steps you have to take if you want to move through your house without the lights? Is the switch always "on" and the sensor then turns on the light and you just have to turn the switch "off"? And then if the switch is turned "off" will you need to first turn it "on" again in order to make the automation work properly again?

Smart switches are always on to the switch , which then switches the light on/off either via buttons or the network. So it doesn't matter if you manually turn them on/off - they remain network accessible. This is why they're strictly nicer to use than smart bulbs. Even if you want smart bulbs, you'll probably want smart switches too, but if you have smart switches you only really get a benefit from smart bulbs if you…

> you only really get a benefit from smart bulbs if you want to control the colour of the light

You can control brightness of a smart bulb. Not all dumb bulbs are dimmer-friendly. That said, you can always replace those with dimmable dumb ones.

Re: I built an offline smart home

#152

Earlier quoted context omitted.

Thats not Hue’s fault though, your friend should be better at planning and implementing their Hue implementation. Hue bulbs default to “normal on” when power is restored, and also work with sensors, dimmers, and switches.

No, it's fundamentally impossible. He wants to be able to control his lights from his phone, and that's impossible if the lights have no power. The only way for his lights to have power at all times is for the light switch to be on, but I don't want that when I'm sleeping. Therefore, the switch must be on but the light off, so to turn the light on, I need an app.

No neutral smart switches exist. The nice ones use a capacitor behind the light fixture to trickle enough current for the switch.

There's actually no particular reason you couldn't do it the other way round I suspect but I've not heard of anyone implementing it.

Re: I built an offline smart home

#153
post #122
post #34

Earlier quoted context omitted.

> Agreed, but Alexa fails hard if you have no internet It also fails hard with Internet. I’m sick of being on my living room asking Alexa to turn on the light just to get the bedroom light turned on (each room has its own Alexa/Light pair, so there should be no confusion).

If they're assigned to the same room as the lights there shouldn't be, but check in the (awful) app which of your Alexa devices picked up your voice. Walls reflect sound, and things like curtains muffle it, and you might be surprised which device picks up your voice best in different positions. I had to adjust placement etc. to prevent the Alexa in my son's room from regularly picking up my commands from my room bett…

Yes, I had interference at one moment and moved the devices to the farthest points on each room, so that doesn't happen now, but the problem I mentioned before (turning on the wrong light) still happens.

I forgot to say that the problem mostly occurs when I ask Alexa to set the (unspecified, but hopefully the one belonging to the room the corresponding Echo is) light's brightness to X. Sometimes, in that case, it turns on the light from the other room.

I didn't check the audio transcript when this happens though, that's a good one, thanks.

Re: I built an offline smart home

#154
post #122

Earlier quoted context omitted.

If they're assigned to the same room as the lights there shouldn't be, but check in the (awful) app which of your Alexa devices picked up your voice. Walls reflect sound, and things like curtains muffle it, and you might be surprised which device picks up your voice best in different positions. I had to adjust placement etc. to prevent the Alexa in my son's room from regularly picking up my commands from my room bett…

Yes, I had interference at one moment and moved the devices to the farthest points on each room, so that doesn't happen now, but the problem I mentioned before (turning on the wrong light) still happens. I forgot to say that the problem mostly occurs when I ask Alexa to set the (unspecified, but hopefully the one belonging to the room the corresponding Echo is) light's brightness to X. Sometimes, in that case, it tur…

That's an odd one. Not seen that. Lots of aggravating little quirks, though. Annoyingly so many of them would be trivially addressable by users if we just had a way of adding simple rules.

E.g. "this request on this device refers to X", or more generally "when you hear X it probably means Y".

As an alternative, I've spent more time than I should passive-aggressively repeating the same instruction and then flagging every single failure in the app. I have no idea if it makes any difference, or if the things I've done that with that have improved were due to general improvements to their voice recognition, but it gives me some limited satisfaction (I now wonder what the financial tradeoff is of fixing the reported issues vs. any benefit potentially gained by just having the reporting function but sending it straight to the bin to make people feel like they're being heard...)

One of my goals for this year is to experiment with HomeAssistant or similar to at least start to replace or supplement my Alexas.

Re: I built an offline smart home

#155
post #120

Earlier quoted context omitted.

Smart switches are always on to the switch , which then switches the light on/off either via buttons or the network. So it doesn't matter if you manually turn them on/off - they remain network accessible. This is why they're strictly nicer to use than smart bulbs. Even if you want smart bulbs, you'll probably want smart switches too, but if you have smart switches you only really get a benefit from smart bulbs if you…

> you only really get a benefit from smart bulbs if you want to control the colour of the light You can control brightness of a smart bulb. Not all dumb bulbs are dimmer-friendly. That said, you can always replace those with dimmable dumb ones.

That's true, I guess I should have expressed that slightly differently - my point was that the only capability you can't get from "just" a smart switch irrespective of your choice of bulbs is changing colour.

But also, smart bulbs with a dumb dimmer is the worst combination, so if upgrading somewhere, if it has dimmer switches that's an extra reason to replace the switch too, irrespective of whether you also replace the bulbs...

Re: I built an offline smart home

#156

Earlier quoted context omitted.

Yeah the jokes tend to be about how little concern there is for quality control or things breaking. I'm a fan of "move fast and break things" for getting things started and new projects but I'm not sure why that mentality continues into mature products. I think a lot of software dev teams would benefit from referencing the TRL system[0]. I did work as an engineer in a aerospace startup and you definitely move fast an…

Does it not matter what the consequences of failure are to determine how much effort should be put into handling failure?

Of course it matters. That’s the “end effect” portion of any proper risk management process. The severity and likelihood of various end effects drives improvement in the design, creation, and usage of the item.

Re: I built an offline smart home

#157

Earlier quoted context omitted.

I so badly want an old school flippy light switch that physically flips when toggled by the smart home. IMO that would be the first smart home product strictly superior to its analog counterpart.

A pressable button that toggles the light state achieves the same thing without the need for a motor.

But they don't physically show their state, don't persist across power outages, and are much less satisfying to operate.

Re: I built an offline smart home

#158

Earlier quoted context omitted.

Does it not matter what the consequences of failure are to determine how much effort should be put into handling failure?

Of course it matters. That’s the “end effect” portion of any proper risk management process. The severity and likelihood of various end effects drives improvement in the design, creation, and usage of the item.

But why are people surprised then that in most software the process is not as thorough as in aerospace or elsewhere with physical things.

Most of the time physical devices failing will have much more severe consequences, than a website going down for an hour. In some cases there are APIs which many other critical systems use and frequently those APIs do have much more rigorous process in making sure that they are reliable.

All of that seems regulated by governments and markets to me.

They have promises of reliability and big financial penalties if they were to fail on that reliability.

Re: I built an offline smart home

#159

Earlier quoted context omitted.

A pressable button that toggles the light state achieves the same thing without the need for a motor.

But they don't physically show their state, don't persist across power outages, and are much less satisfying to operate.

I usually look at the lights instead of the switch to determine if the lights are on.

I haven't experienced many outages recently, but I'm fine with lights off until I turn them back on as a default. Its also possible to just have the lights revert back to their previous state as an automation in HA, although it seems like a waste of time to program instead of turning lights back on as you need them.

Re: I built an offline smart home

#160

Earlier quoted context omitted.

Yeah the jokes tend to be about how little concern there is for quality control or things breaking. I'm a fan of "move fast and break things" for getting things started and new projects but I'm not sure why that mentality continues into mature products. I think a lot of software dev teams would benefit from referencing the TRL system[0]. I did work as an engineer in a aerospace startup and you definitely move fast an…

Does it not matter what the consequences of failure are to determine how much effort should be put into handling failure?

That's actually the entire point. There's consequences to however anything fails. Think the skyscraper example. Do you want to explicitly design a failure mode or do you want to be just "let's see what happens?" Which do you think has a worse consequence? Neither are great, but one is definitely better than the other.
Post reply on HN