My comments on Smartthings platform and its future evolution
As you know, for 6 months I was offline, but I was still totally online, pure Smartthings style.
My comments may be wrong, it would be more than that!
Observing the firmware evolutions of the Hub, the activity on github of the stock drivers and the app I will go comment part by part:
Smartthings has bet almost everything on Matter and is determined to be a very, very major player in its development.
Matter is insatiable and consumes more and more resources from the Hubs. The drivers have grown so much that only their engineers and @mocelet understand them, .
I never went into matter develop and Iām not going to go into it, in this you will be the testers and when it all works weāll see. My needs are covered with zigbee and zwave devices.
This consumption of resources has brought the groups of hubs, which are at their limits, and the development of a new Hub.
Smartthings has invested a lot of money in these 5 years to consolidate the groovy migration of the zigbee and zwave platform. Donāt forget it. It has consolidated solid libraries, they cover the mandatory specification, almost all the optional, it does not cover the custom. They are reliable enough, that they allow smartthings and its partners to offer a smart home complement that gives value to their smart samsung appliances and we can automate our custom home and thus not lose users to others platforms.
The changes in the stock drivers code are only to adapt them to the firmware, improving performance and above all freeing up hub resources for Matter evolution.
At this current point I think SmartThings will not take care of integrating Zigbee and Zwave devices. It will have to be their partners and manufacturers who, through easy WWS or with their own developers, make drivers using lua libraries and distribute them on their channels, Aeotec, inovelli, zooz, sonoff and many others. This is already the many case.
For other devices (tuya or with custom features) we will still need community developers. There are almost no more of them, we have not taken good care of them, or not very generous with their work. some could have found work in manufactures integrating devices in smartthings.
It is not a matter of monetizing, a horrible word, it is a matter of solidarity and gratitude. Iām going to give you an example:
Many of you have been generous with me and others developers.
Imagine that one day I say to my wife, who suffers daily from my lack of attention due to my dedication to community Edge drivers, Do you think that tomorrow you and I are going to dinner?. When I go to pay, I tell: This exquisite dinner is given to us by my friends from the smartthings community so that we can enjoy it together. The rest of the night I wonāt tell you, but you can imagine it, hahaha . How simple and comforting on both sides!
You will take care of yourself and maybe you will not lose those that remain.
There are also external developers whose work you should thank: kkoŔsev @Trakker2 of hubitat, what would hubitat be without him. It has helped me a lot with the Tuya black magic and its code on github, thank you
koenk allows us to use his libraries to integrate complex devices EF00 and others, thank you.
The App: One of the best on the market I think and it will continue to improve quarterly, I tell you something not confidential.
A few months ago smartthings finally understood that they needed to improve pre-launch testing and created a new App beta group to test them. They invited users from the community, I was invited to participate, thanks, but I was angry with the world and I turned it down. This is the way, I donāt know why they themselves did not announce it to the users.
In my conclusion, smartthings is more alive than ever and will continue to be a benchmark. I recommend Do not abandon Smartthings, if you need to complement, do so, but do not abandon, you will regret it later
Several platforms will works together to smartthings, I think
I agree, I think many power users will end up combining multiple platforms at least for a year or two while matter catches up with its promises.
Smartthings also remains a very good choice for those who like Samsung smart televisions or smart appliances and want to run without an additional hub. Itās pretty easy to add a few smart lights, a smart lock, and a couple of smart plugs, and still have pretty advanced routines. So thatās another market segment that I think will be happy with smartthings even if they are looking for very different things than the power users who want to run everything locally.
Taking advantage of your answer I am going to give my opinion on Matter:
I consider it one of the biggest scams to users by big tech.
Now I will explain it in detail, in this conversation, I donāt want it to be in any specific one that already exists, they have become so technical and convoluted that I think 90% of users donāt even read them. They are a mixture of self-serving spam by manufacturers to incite their use and create expectations, which we will see if they are fulfilled.
I encourage you to create conversations at a lower technical level where users can inform themselves and make useful decisions about whether to use, not to use or if I need matter now.
After this headline I have to make a point to tell you that the neurosurgeon and his team from the Spanish public health, who repaired my brain in a 5-hour operation on September 1, must have taken the opportunity to do a cleaning of neurons that has left me with an unusual brain activity and a clarity of ideas boiling that I have to share. Thank you Pablo!!!
Some decided that a common platform had to be created, which would save them a lot of costs with minimal industrial competition and also facilitate a more standard, not better, use to users.
They said, whoās going to pay for all this development?: The users, of course!
We are going to make a specification, a part only, all of it is very expensive, that allows us to put the industry to manufacture and sell products and thus finance the project.
We are inundated with targeted advertising with expectations that blind us.
We started buying matter products, so to speak, and at that point we started working for them and they didnāt get the services weāve paid for.
They are putting the carrot in front of us so that we do not stop, matter 1.0, 1.1, 1.2, 1.3
An industrial book scam!!
Will it ever work? How will it work? I am going to give my opinion with little technicality, which I do not know in depth either
Matter will work in a very standard way in general and very personalized and different depending on what application or final platform we want to use
Samsung, Philips, Apple and other prestigious brands are not going to give up on making money and users by offering more personalized features.
The specification allows it and the specification who ultimately executes it is the end user when they use their application to turn on or automate a simple switching on of a light regulator.
The specification says how the ignition should be done and the device knows it too. If you just come in a command on and 50%, said and done.
But they know that there are many users who want it to turn on at a specific time, to start at 20% and reach %50% in 10 seconds, for example.
This is what manufacturers and their applications and hardware are going to invest in.
The software and hardware that allows you to customize the devices will be developed on your specific platforms, smartthings, hue, nanoleaf, ā¦ They will invest in this to gain market.
The end user will decide whether to settle for standard operation or want to use specific platforms that allow them to meet their needs. You can even use both at the same time.
That point has not yet arrived and will take time, I think.
Everyone must decide if they want or need to embark on installing matter devices in their home now. I wonāt, but I appreciate that there are intrepid users who pay and suffer in their flesh the development of matter that we will all enjoy in the future
Ā”Ā”Ā”Ā”Ā”I have to stop, my neurons are overheating!!!
Something that particularly concerns me at the moment is the hub groups.
In the last few weeks I have had a couple of incidents where Zigbee has gone a bit odd. The details are a bit fuzzy but basically I have had situations where certain Zigbee sensors report offline and all my Sonoff Zigbee buttons have stopped working, yet the app can still control Zigbee smart plugs. Third Reality Zigbee buttons either still work or recover more quickly once I have rebooted hubs. It is a mess and I donāt have a scooby what is going on.
So where do the hub groups come in? Well the integrity of the hub groups seems to be determined to a significant extent by Zigbee connectivity. With the Zigbee playing up as above I am told that the primary cannot connect to the secondaries and I should perhaps move them closer together. They are less than six feet apart as it is. I say ātoldā but the reality is I donāt get told this at all. I donāt even get any indication on the Home Insights dashboard. It is only when I go looking at the hub groups that I discover that there is an issue.
If you have the hub group set to automatically failover you are led to believe that you get notified when there is an issue and you then get the opportunity to decide whether to continue with a hub backup or not. You donāt and you donāt. Instead a hub backup silently happens and you just have to hope that it works smoothly. Invariably it didnāt for me because the hub backup results in an āaddedā lifecycle for each Edge device on the new hub and no one warned developers that this could happen and so any dutifully created synthetic events to keep the mobile app happy were emitted again. Particularly a problem when you are using virtual sensors. Also matter devices can take a ridiculously long time to come back online after a hub backup. It really isnāt something you want happening unsupervised. Oh and if you have a preferred primary hub the idea is that when that recovers it will take over again. Well again that is a silent thing and for me despite suggestions that it was imminent for hours on end it only ever happened once.
There is also the question over why the primary and secondaries supposedly being out of contact merits a failover anyway. Whenever I had this happen there was nothing wrong with my primary in the first place and the best course of action would have been to remove the secondaries.
I have a hub group purely because that is the only way I can get more than one of my V3 hubs on the same thread network and those are the only useful TBRs I have. I donāt need those hubs routing Zigbee or Z-Wave and Iād rather they didnāt. I donāt want them to have to be close together either. I want them spread out.
It worries me that Hub Groups seem to have become mainstream.
I bought a v3 hub a few months ago to test with hub groups. Moving devices from the floor 2. I donāt need any more hubs.
My V3 placed on the 3rd floor of 3 and maintains a very robust zigbee and zwave network I have no problems with offline devices.
I started reading how hub groups work and itās all so poorly documented and obscure, that I said: Iām not going to put the home automation installation at the service of smartthings, so that they can develop something they need because I think matter is insatiable with the resources of the current hubs. They have to test and make sure that the hub groups work well before offering it to users. We canāt be used as beta testers without us knowing that we are.
I donāt have any data to comment on how hub groups work and their problems, I read in the forum that problems exist.
My simple advice if you have memory limit problems, which create failures, then try to reduce the hubās memory usage and if you canāt use the hub groups, you have no other solution with smartthings.
Iām going to comment on what I do to improve memory usage and try to avoid using hub groups:
I have 100 devices, zigbee, zwave and virtual with 24 drivers Mc installed, 155 routines and scenes, no ruleAPI
They control all heating, cooling, lights, energy meters that report every 1 minute for solar generation control, which automates the use of surpluses daily, security, contact, vibration, motion, smoke and water valve sensors.
The memory limit is at minimum (OK). 1/3 of the total memory?
Iām sure it could handle the maximum, 300 devices without problems.
Most are cheap zigbee devices, sonoff, tuya, moesā¦ I always buy the same models. First I buy one and if it works I buy more of the same. That ensures that I can use a single driver for multiple devices. 1 driver for a device is a waste of memory. Manufacturers must be required to group several models in their distributed drivers.
this o do by Aeotec, inovelli, zooz from the beginning of edge with their own edge developers that are present always in our community.
For others, as sonoff, friend and others this must be done by buying devices only from those who comply and telling them so.
My zwave devices are fibaro and aeotec.
I only use my Mc controllers, which allow me a lot of custom features and settings:
Customize reporting times between 300 and 3600 sec. I have them between 1800s and 3000sec for the battery ones
Between 300sec and 1200 sec switches, plugs and bulbs and remotes. I have them between 900 sec and 1200sec, so they donāt all report at the same time.
My zigbee network is silent, I can see it in the log cat
Stock drivers report every 300sec normally and the net looks like a cricket cage.
If the Mc driver uses subdrivers, the lazy_load_if_possible function is always used, which ensures that only subdrivers that use a device are loaded into memory.
Stock drivers only use the lazy_load_if_possible function in 2 drivers, zigbee-switch and zwave-switch donāt understand!
If you donāt need matter devices, donāt buy them, they are constantly evolving and you are buying possible prototypes, which many will end up in the trash when matter finally works.
I know that not all the recommendations can be implemented by everyone, but some can.
Hi @posborne I hope youāre okay and @SmartThings@nayelyz Please could you do this for the team? I didnāt realize they donāt read the forum.
The lazy_load_if_possible feature was implemented in the v9,(52.x) libraries
We are in the v15 beta libraries (58.x beta)
This function frees up space in the memory of the hubs, loading in memory only the subdrivers that have a device installed, as you acknowledged in your posts of feb 2024.
To date, if Iām not mistaken, only zigbee-switch and zwave-switch have it implemented in the published stock drivers.
There are many users with memory limit problems that generate many failures in their home automation installation.
Please, Could you or another team member tell us the reasons for the non-implementation of this feature on many other zigbee and zwave drivers, which have many subdrivers and control many different devices?
Maybe implementing this feature could alleviate some of the memory issues of current hubs, especially v3, but we donāt know.
I have implemented this feature in all of my drivers Mc since March 2024 without problems in more of the 31000 user if my shared channel and my self.
Samsung and smartthings are a great industry and home automation platform, I am convinced that it will continue to improve and with its cutting-edge energy-saving technologies they will contribute to the ecological transition, automating and making our homes and industry more efficient with renewable energies and keeping alive this planet that we have to leave to our children.
Thank you and much encouragement to keep improving!
Hi, Mariano, following up regarding this specific topic, this is the feedback:
Lazy loading has a greater impact on drivers that contain several sub-drivers, so those that benefit the most had it implemented first. Other drivers with fewer sub-drivers would benefit considerably less, so implementing it in those drivers is less of a priority than other upcoming improvements for hub memory that the team is hard at work on, and we want to get released to you all as soon as possible.
Thanks for the answers.
It seems logical to me to prioritize resources and actions to achieve the best results.
Perhaps you could study the memory impact of implementing it one driver, with many subdrivers and many different devices widely used in all home automation installations: Zigbee Motion sensor (15 subdriver)
.
. How simple and comforting on both sides!
