Can we revisit the move to qt-openzwave?

paulstronaut · January 1, 2021, 6:21pm

I would love if we could start an open and civil discussion about the future of Z-Wave in Home Assistant.

Here is my current understanding of the state of things:

Context:

In November 2019, it was announced that HA would be deprecating the existing Z-Wave integration and moving to the addon for QT-OpenZWave. There was a lot of excitement.

Over a year has passed since then. Initially, there was a lot of movement on fixing bugs and making the integration more usable.

In the past 6 months or so, the project has become stagnant.

Current state:

The initial “Future of Z-Wave in HA” thread has turned very sour. People are struggling with many issues in the beta (too many to list – see the thread linked or search the forums). They apparently have been reported here and to GitHub, but those who are in control of the addon, understand how it works, and are able to merge fixes are somewhat MIA. This, while understandable because it’s all volunteer-based, is a hard pill to swallow for a lot of people who rely on Z-Wave working.

There have been people bringing up alternatives only to be shut down (somewhat rudely, tbh) without any chance of discussion. I even tried to ask some direct questions about the decision with qt-openzwave, where would be a better place to discuss, etc, but received no responses.

Going forward:

How do we move forward? Those who really know the qt-openzwave code in and out seem to not have interest in helping. Those who understand new projects want to help and are offering their hard work, but are basically being told that their efforts are wasted because the Home Assistant crew does not want to even consider an alternative.

petro · January 1, 2021, 7:22pm

What are you talking about “has gone stagnant”? I just updated from a 6 month old version and it had a ton of updates, basically fixing all the issues I had.

paulstronaut · January 1, 2021, 7:26pm

Here’s one such post discussing lack of updates The Future of Z-Wave in HA - QT-OpenZWave

petro · January 1, 2021, 7:35pm

Honestly I’m confused as to what the issue is? Having it zwavejs2mqtt as a community addon instead of HA maintained addon isn’t really the end of the world…

Also, fwiw, I run qt-openzwave with 70 devices and have no issues.

blhoward2 · January 1, 2021, 7:41pm

There have only been a handful of merges since July if you look at github. And those all originated from PRs that sat for months before being committed. The maintainer, fishawaldo, has only taken an action in any project a handful of times since July. The issues list continues to grow and he hasn’t even really commented on them since July let alone fixed (or committed PRs fixing) any of them. It’s stagnant.

This is the sole maintainer’s activity across all projects:

A critical bug causing the entire thing to crash and to take the processor to 100% due to an MQTT timeout value has been open since July. A PR to change the value has been pending for some time, as an example.

petro · January 1, 2021, 7:43pm

Ok, then use the new js one. Why does HA have to manage the addon?

blhoward2 · January 1, 2021, 7:45pm

I think the issue is really that if you look on here, or the HA thread on Reddit, tons of people are trying the beta because of the announcement pushing people to do so. But since it is still broken in several ways they waste a ton of time and then give up frustrated. I’m talking hours of renaming entities only to have to undo it all. HA is actively pushing people to a project with critical bugs and no pathway to fix them at the moment.

I’m not sure that HA does have to manage the addon, but the messaging is confusing. As recently as the conference qt-openzwave is being touted as the pathway. In my humble opinion the project needs to put out messaging putting a pause until development resumes so that people know.

petro · January 1, 2021, 7:52pm

It’s still in beta though. It even has the name. It’s enter at your own risk for a reason. How about we just wait until the project is announced dead instead of speculating.

blhoward2 · January 1, 2021, 7:53pm

Frankly, someone needs to reach out to fishwaldo and see if he still intends to development qt-openzwave at all. A lot of plans are being made around something and unless the project leads have inside insight (which they may) it seems like it is predicated on something that has stopped happening.

Like you, I’m not clear that people realize how stagnant development has become. Or the number of critical bugs being experienced by a large number of users. Mine worked fine with 100 devices for months. Then it didn’t despite nothing changing, not even updates.

petro · January 1, 2021, 7:58pm

His last correspondence on the forums indicates he’s ran into personal issues but he plans on getting back into it.

All the more reason to just be patient. I’ll attempt to reach out to him but with all the world events going on, I’m not surprised some people have to take breaks.

blhoward2 · January 1, 2021, 7:59pm

Yes, but that post was in like September wasn’t it? Obviously he has done a lot of great work over the years. He owes us nothing. I actually worry about him because of that post and his sudden absence. I just don’t want to see people frustrated, and development of that portion of HA stalling for a significant period of time, if he isn’t coming back or isn’t coming back anytime soon.

petro · January 1, 2021, 8:02pm

October. But that shouldn’t matter. You don’t know what personal issues anyone is facing. Again, everything mentioned is just speculation.

blhoward2 · January 1, 2021, 8:03pm

Not sure if you saw my edit. I’m not upset at all that he’s taking a break. My only point is that it has now been a significant period of time and so it seems time to checkin to see what his plans are. I appreciate you saying that you’ll reach out to him.

It makes little sense to keep directing people to the project if he’s going to be gone for a while or is stopping development. They does no one any good, including him.

I’ve never met him but I’m actually worried about him some. I hope that he is alright.

petro · January 1, 2021, 8:15pm

Like I said above, it works great for me without any critical crashes. I’ll still promote it because it works great imo. But a community is going to promote what it likes… if people are promoting it, then it must be working pretty well.

blhoward2 · January 1, 2021, 8:25pm

I’m not sure that I follow this. If it works well for a few people at the top of the project and they use their relative power within the project to encourage its adoption, that is not the same as widespread community support. If you search the forums the wider opinion seems to be the opposite. Almost all of the recent posts are discussing major issues or speculating whether development will restart. There seems to be a disconnect between the main contributors and the average experience, which is what is leading to this very type of thread.

petro · January 1, 2021, 8:26pm

Also, I’m pretty sure he’s still active, 15 days ago he made commits to the main openzwave project.

petro · January 1, 2021, 8:28pm

There’s more than just a few people who use qt-openzwave. Just like any forum, ‘issue’ posts are misleading because it seems like they are widespread. But in reality, 15 people having issues on a forum with 100,000 people does not mean “everyone is having issues and only a few aren’t”.

blhoward2 · January 1, 2021, 8:28pm

That is a fair point. I don’t dispute that squeaky wheels make more noise. I don’t think it is 15 people though. And I’m not sure how many people are actually using it day-to-day.

petro · January 1, 2021, 8:33pm

There are too many to count using it day to day. I spend a ton of time on the forums, github, discord, and reddit helping people with these issues.

paulstronaut · January 1, 2021, 8:37pm

If I could close this topic, I honestly would. This is already devolving outside of a good discussion and into just two people chatting, not really getting anywhere. There are obviously some really big issues, and the fact that it was also announced that we should expect the qt-openzwave implementation to move out of beta within the next three months is very concerning, given the issues that have been reported.

I was hoping that we could gain more insight into what the actual forward plan is. We can’t rely on a project that has a single core maintainer and we should be able to agree on what we feel is the most stable and easy to maintain piece going forward – that HA can recommend as an “official” integration. Or just stop recommending “official” at all.

@petro you’re being really civil, but as a moderator, your short responses and way you dismiss issues that we want to bring up is part of the issue that I’m trying to point out with this thread. It really sucks as a community to run into things like this and just be shunned away.

Maybe we can sit back and let more people from the community put up their points?