I'm unhappy with the removal of GPIO

This should be exactly the reason why people should not use GPIOs in a constaly updating system that they could loose at any moment becuase of an update, a corrupted SD card etc. Especially when we are talking about an alarm system for your house. Those should really be standalone, which is exactly why something like an ESP device would work so much better there.

1 Like

My god man, just stop with your threats of stopping supporting HA, nobody is forcing you. You have said the same comments multiple time in other post. Just do it but we dont need to know

2 Likes

I fully agree. Donā€™t fix what isnā€™t broken. I rolled back to the previous core for this. Plus GPIO issue.

there is no rp_gpio issue in this release, its deprecation warning. You have depending on the integration 2 or 4 months to find alternatives before it becomes an issue for you, so rolling back wont magically make that disappear its inevitable.

1 Like

I rolled back for the binary sensor state change. I really donā€™t understand why that was doneā€¦ On top of it it doesnā€™t work. I get a unknown state where state is well known ? Why?

And who are you again? :thinking: Canā€™t find you on GitHub or Discord, just here, ā€œprivatelyā€. :roll_eyes:

(this is the polite response to your comment)

2 Likes

you obviusly didnt look very hard

3 Likes

@thecode, thanks for chipping in. I donā€™t particularly like the way this deprecation was done (not your doing) but the sentiment about direct hardware are IMHO quite right. There is a place though for rpi direct control of itā€™s own cooling fan, even if that is an addon to the rpi.

I came to the conclusion not long after I started on my HA journey that it might be necessary to be able to run up another instance pretty quickly. The forum was (and still is) full of tales of SD card woe. I figured no point in tying my home automation to just one pi. Running sensors or controllers to the hardware of the pi seemed a mistaken approach.

As I said, use of gpio for pi hardware functions, like cooling fans, makes perfect sense, and there will be many people who are eternally grateful to you.

If you need any hardware for testing, eg 1 wire devices, sing out. I am sure there are plenty of people willing to donate to keep this stuff alive.

Legacy Nest - #7 by allenporter has more details on works with nest. The issue with nest is maintaining two integrations within a single integration is pretty difficult (otherwise I wouldnā€™t care) and less about number of legacy users vs sdm users. Itā€™s a fantastic candidate to move to a custom component.

1 Like

That has been my reasoning also.
But ā€œtheyā€ fear the extra layer of software and WiFi.

Unfortunately this is a recurrent issue , so may be the solution is to force people to click on a button ā€œI have reviewed the release notesā€ before the system starts the upgrade.

You often say that. This implies people are ignorant. I do not assume that this is intentional. But it gives a lectured and stupid feeling. Ok, I also get annoyed sometimes and think to myself ā€œthat was only asked the other dayā€. But Iā€™m reading daily in this forums. Although Iā€™ve never seen this link to the architecture site before.
In fact, after only a few hours, at least 100 new topics have already been posted. Postings are simply lost in the face of the mass of new topics. And as the way is with algorithms, the search function doesnā€™t sometimes help very much (especially for non-native speakers).

1 Like

Itā€™s in the Breaking Changes of the current release notes.

Itā€™s posted elsewhere as well.

It does not imply people are ignorant. Iā€™m implying that Iā€™m sick of saying the same thing over and over again every release. Let me give you a short timeline of any HA release:

  1. Some change happens that a small subset of users donā€™t like
  2. They start complaining that they didnā€™t have a voice in the discussion.
  3. Shortly after, it comes to light that there was an open discussion with a link, guess what, itā€™s on GitHub.
  4. People then complain that they shouldnā€™t have to join GitHub to be involved in the discussion or that it should have been more public (what???)
  5. Itā€™s then explained that GitHub is where all the code is addedā€¦ so it only makes sense to discuss it there. You know, to have a chain of history attached to each discussion. (Holy moly, makes sense).
  6. People are still upset regardless.

Myself and other moderators have to deal with this every single release. Normally this wouldnā€™t bother anyone, but as it turns out, itā€™s usually the same people who are upset. And it turns out that these same people arenā€™t affected by this change, they are just joining in the argument.

How would you react to situations like that constantly? Would you be annoyed or content? Iā€™m betting youā€™d be annoyed.


Nothing in this software is a closed decision. Everyone can chime in on any discussion on GitHub about the direction of the code. Keep in mind, you still may not get something you want out of the discussion. In the end, the decision comes down to the code owners/maintainer, however you can still make your case politely.

5 Likes

Yep, very open when discussion about it is at githubā€¦ again, not for us nondevs. Conversations have to be brought to blog level if there is any desire to get even a little bit mainstream.

Btw, this thread is for complaining about removal of the gpioā€¦ not for complaining about complaining :wink:

2 Likes

Another alternative I havenā€™t seen mentioned in this thread is a GPIO microcontroller connected by USB serial to the RPi Server. This is possible for instance with a board running Firmata. Something I just learned recently.

The boards are cheap and there is no relying on the network. I read some people mention alarm systems running via Rpi GPIO, I have to agree with others that running alarm systems as a standalone capable ESPhome seems to be even a better idea.

So next to the alternatives as, among others, the Rpi GPIO HACS integration, (wired/ethernet) ESPHome, or MQTT IO, this can be considered as an alternative too.

I think thatā€™s to late for affected people to join the discussion :wink:

@Petro:
When Iā€™m annoyed I think my piece, but I wouldnā€™t put it that way (to answer your - maybe just rhetoric - question). My language skills are really poor but ā€žno one takes part ā€œ means to me ā€žpeople donā€™t want toā€œ = ignore.

To be clear. I understand you, thatā€™s not the issue. Itā€™s about dealing with human behavior.
I donā€™t want to offend anyone (at least anyone who has made acceptable comments), and I donā€™t do myself any favours by getting upset with others. People will get always upset with certain topics, itā€™s typical for internet forums. The release thread is anyway hard to read because of the jumbled posts. You can let them tear each other apart (stupid for forum culture), close the topic (stupid for everyone else), or separate the discussion. Anything else will lead to trouble (just my two cents, not an instruction :wink: ).

Just a small note on point 4. Once, I donā€™t even know what it was anymore, I published that users should keep track of developments and github. The answer was: there is no obligation to monitor this. And that will discourage you from further discussion.

1 Like

I assume you are pulling my leg in jest because the deprecation decision was discussed, in this forum, as far back as December. Thatā€™s well over a month before the deprecation was formally announced. In addition, the actual removal of the integration is months away.

Yep, those might be good for some. Iā€™m checking out some industrial dinrail ethernet esp32 boards with isolated IOs.

I am not running alarm system on rpi gpio, just using it for some controls and automations. Alarm is separate and stand alone. Gpio also used for shutdown/ups and extra layer for locks.

Have to start thinking HA also as just another module and convert as much as possible away from it. Already took out all basic automations that i ran with my Shelly gear and automated them on Shelly interface.

2 Likes