Disappointed in the new release - why focus on a cloud platform?

Oh, definitely. No question cloud is evil as far as home automation goes.
Internet down? Screwed. API deprecated? Screwed. Vendor deciding to charge for API? Screwed.

Which leads back to my very first comment: DON’T buy cloud-enabled devices…

3 Likes

284 poll vs. 99 push.
But that doesn’t really matter. As soon as you go to cloud either way, you’re leaking private information.

I feel sorry for 2 groups of people:

  1. Those who bought a whole lot of cloud devices because smart homes were cool. Eventually quite a few realise they are not all they were cracked up to be and are stuck in a sunk cost situation with an inferior solution. They simply made a decision that in hindsight they regret.

  2. Those who bought something like Tuya in the promise of future local control.

3 Likes

Which brings me back to an old discussion I had, notably with @aidbish : HA (and home automation in general) DOES require knowledge and study :wink:

2 Likes

touché…

1 Like

Don’t be that hard, nickrout. As we all have already learned from the Tuya marketing staff:

… comparing apples with pears to convince people that nowadays local control is not a must anymore but can safely be neglected :rofl: :rofl: :rofl:

1 Like

Wow. This topic kind of exploded. With all the discussion above, I want to clarify something.

Yes, a smart home needs cloud these days. I’m not against that. I’m also not against the working with vendors to create new integrations. What does bug/worry me is the following:

Normally standards for getting an integration added to HA are really, really high. Go through the pull requests and you will see that very often developers are forced to split up changes or to do things differently because their work does not meet HA standards. I’m amazed that the Tuya integration could pass these checks.

The way you have to configure this integration is convoluted to say the least. Look here and on Reddit and you will find plenty of HA users who have issues in getting things set up.

But my main issue is this: why give this so much attention? Why give Tuya such a forum while there are so many other really good integrations? Especially because the quality of the integration is really poor. This just reflects poorly on the judgement of the HA team and indeed, makes many people question if there are no alternative motives behind this change.

The silence of the HA and Tuya teams in this matter is not making it any better either…

4 Likes

They do this ‘spotlight’ stuff for every ‘large’ integration that affects many users. Ya’ll are grasping at straws here. This is a simple release of a new integration that happens to affect 14% of users, this integration happens to be one of the ‘big’ items released this version. The old tuya integration was cloud_poll and the new one is cloud_push. It only makes sense to rewrite when it’s a completely new API, and the old cloud_poll integration had no code owners. And yes, the integration ticks all the home assistant requirements. You’re welcome to look at the tests for each PR, you’ll notice every test passes. You can even compare the integration against the ADR’s.

Are you all forgetting that this same push occurred in January for zwave? What about the past pushes for other systems? Smartthings, Unif… Etc. All the history is there for all of these. There’s no mad conspiracy.

You’re all welcome to join discord. There you can see where all the correspondence for everything goes down. There’s roughly 30 channels dedicated to different parts of HA, including ~15 dev channels. All open for everyone to see.

1 Like

But the lack of support has been considerably noticeable for this new integration.

While a lot of stuff can be gleaned from discord, there were meeting between the devs and Tuya that people on discord would not be privy to including zoom meetings

Cause it’s one dude who did most of the changes. :man_shrugging: It’s not like zwave js where there was ~15 of us

Look, I’m willing to believe that there are no nefarious intentions behind this integration. But the fact that it has gotten so many people to react doe give some indication that something about this one is different, at least in the perception of people. Maybe it Is because the loss of functionality, maybe because it is cloud, maybe because of all the promises, maybe because Tuya is a big player. Probably a combination of the above.

In any case, I do feel that the way this was handled is far from ideal. A community is based on trust and whatever breaks this trust risks breaking the community, regardless of how good the intentions behind it are. And nobody wants that.

3 Likes

False equivalence; Home Assistant isn’t a car, it’s an open-source software project, and unlike a Mercedes it’s free (and all that that implies). But if we use that flawed comparison, Mercedes makes Bosch fix problems with Bosch’s parts.

Like all contributors of integrations to the Home Assistant project, Tuya knows their own code best. Until someone else understands it, they are best suited to correct/enhance it.

They can take responsibility for mistakenly believing Tuya deserved the marketing it received. However, Tuya’s devs remains responsible for the integration’s code, the extent of its functionality, and the timing of its release … and fixing its bugs (especially if it requires enhancing/correcting their API).

If we are making an exhaustive inquiry into who should shoulder responsibility, let’s not overlook the Tuya users who did not participate in beta testing.

1 Like

Broadly agree but…

…surely HA control the timing of when core integrations make it into the erm… core?

Taken from the Why Home Assistant doesn’t have an external API for integrations May 13th, 2021.

The second line is key - they didn’t make the manufacturer work harder for the benefit of HA and it’s users.

2 Likes

Disagreed, I don’t see a flawed comparison but at least I just learned that Home Assistant isn’t a car. Thank you for the enlightement. Anyway, your interpretation might be different in accordance to another cultural background on the latter.

HA should make Tuya fix problems with Tuya’s parts inside it’s (HA’s) core.

There we are again.

They can’t make a contributor fix it because they have no leverage other than the extreme option of excluding it from official releases. The current situation is far from the ‘nuclear option’ and Tuya is likely to fix the issues eventually.

Even if the integration were excluded, it’s a low-yield nuke because it would continue to be available as a custom integration. The only way a contributor can be coerced to fix bugs, is if maintaining ‘official integration’ status is important to them. Time will tell what’s important, or not, for Tuya.

4 Likes

In this shuffling with arguments all are forgetting about HA users. the fact they have been painfully impacted is out of discussion. And responsibility for this goes to HA leadership.

Don’t know processes established in HA project but it’s not the first time that some change hits users hard. This time the issue is getting more friction because blaming external company is more acceptable on this forum than blaming HA developers, regardless the resulting issues are very similar. But as in other cases there is no any responsibility nor feedback nor support from devs (neither tuya nor ha).
Tuya v2 is not the first half-baked integration rolled out being previously advertised loudly (incl replacement of previous solutions). The biggest fail is, that previous Tuya integration has been removed which made things even worse. But at first such an integration shouldn’t be allowed to be merged at all. Someone should start consider HA users just as regular users who use HA for control their homes (not another team of beta testers). We are lacking quality of HA (don’t argue that there are beta stage everyone can help with - this is not the point)

To me, HA misses strategy for making this product more reliable. I mean more stable, requiring less maintenance, guaranteed no unnecessary outages, with clear strategy for enrollments and decommissioning.

7 Likes

Then I think the ADRs and the approval process need to be looked at.

If the ADRs allow an integration to replace an existing one and in the process deprecates features, even somewhat temporarily, “every test passes” shouldn’t be possible.

A replacement integration should be better than the one it replaces, be easier to setup and use in the process. This is certainly not the case with Tuya. The setup process is complex, documentation was missing steps and overall provides less device support all for the seemingly singular benefit of cloud_push.

I know I’ve already linked this above, but it is from only 5 months ago. HA has, or at least had, a core belief of local control and yet the HA devs are side by side on a live stream for 30mins to launch a knowingly cloud only solution, and celebrating it.

Thanks for linking that.