That’s correct, there is no “direct link” between the foundation and the “development” part of openHAB.
Here’s a proposal that’s technically quite feasible and might satisfy the “helping first time users get started” issues.
We could add a “Help Sideber” analogous to the “Developer Sidebar” that includes things like
- a basic step-by-step intro with links to the relevant OH pages and Getting Started Tutorial pages
- Simple FAQ section
- Glossary section to help with the terminology
- …other things?
Here’s are some mockups of what I’m thinking:
9 Likes
I like the concept!
Some concerns I have is that the sidebar can only appear if the screen is above a certain width assuming it follows the developer sidebar restrictions. I’d be cautious about putting something needed by new users in a UI widget that can’t always appear on every screen. Maybe a separate page opens to another tab?
I also have concerns with making it hidden by default which I presume a sidebar like this would be. Perhaps we can show it by default after that first run of the wizard.
Finally, I think something like this needs an icon to show, not just a key combo. Maybe we can put a permanent icon in one of the corners that only shows for the admin users to bring up the sidebar.
If we make this hard to find, the problem becomes that end users won’t know it’s there.
2 Likes
This is a good point. The reason that I think the sidebar makes a better option is that it can be open next to whatever page the user is working on. It’s far less useful (or even more annoying) if you have to navigate back and forth constantly.
I haven’t looked into it, but there’s probably a way to set it so that in narrow screens it just opens over the whole screen like the left panel does, or even is restricted to the bottom half of the screen. I’m not 100% certain what all the options are there.
I was thinking along much the same lines. I suspect it’s not too difficult to have this open by default if, for example, the overview page is empty and showing the default new page screen or something like that.
100% agree. I think it’s a no brainer to add another floating button across from the ‘+’ button on most of the setting pages. The only issue then is what about when the user is viewing one of their user-defined pages. I don’t think we want to drop a floating button right on top of whatever else is going on for those pages.
Certainly adding a link in the left panel that opens up the help sidebar is trivial. Or separate the “Help & About” into two links: “Help” which opens the panel and “About” which goes to the info page.
3 Likes
I have a production openHAB 4 system, an openHAB 4 lab system and a Home Assistant installation. All in virtual machines. Today I recommended Home Assistant to a work colleague saying, “This is easier for you, openHAB is better, but too complex for beginners.”
Home Assistant comes with its operating system HAOS as ready-to-load VMs. In the colleague’s case, all you had to do was import and launch the OVA in his Synology in Virtual Machine Manager.
On to the first launch. Short questions about location, units, quickly create a user account (real user accounts with rights) and then the first devices found in the network are already detected. You only have to nod off what you want to have and still answer the questions about rooms in which the devices are located.
To stay in the openHAB language, many bindings (integrations) are automatically selected because the devices are detected. If you confirm the usage, the items (entities) are created directly. The names are logically composed of the device name and the associated function.
In the case of devices such as routers from AVM or Hue, rooms are taken over as rooms and only need to be nodded off. This allows a quick start and a good feeling.
The automations are also essentially clickable together by mouse, are stored in YAML and can be edited that way as well.
The makers of Home Assistant are striving to make more and more configurable via the UI, YAML as a configuration path is becoming less and less. A lot of work is done on the UI to make it modern and usable.
In detail, however, there are already quite a few problems, for which I did not want to switch at the moment. But in order not to get blinders, the look in the other community is helpful.
But apropos community. I found the openHAB community in recent years much nicer and more helpful. A clear plus for openHAB. However, I must also say that I also think that openHAB is too little in the focus of the public. The computer magazine c’t from Heise Verlag in Germany was mentioned here. I like to watch their YouTube videos and listen to their podcast. When something is reported on the topic of smart home, openHAB is factually never mentioned. Node-RED and Home Assistant are clearly leading the way. Even in their magazine MAKE.
From a beginner’s point of view, I’m not surprised. I once tried to set up a KNX actuator via UI in my lab version (I’ve been working with files since I got into openHAB). Now I knew the connections, but a beginner certainly does not find himself easily pure. Even if he understands what a binding is, what he wants to address, the binding is installed and the device is recognized by openHAB, happens: nothing. You have to create items, link them and if you just click, you have items with random names that do not reveal their content or function and unfortunately cannot be renamed. Being able to change things is one of the most important functions for me in a system, and unfortunately openHAB fails there with its UI. With Home Assistant, an entity can be renamed at any time without destroying the system.
This perhaps as a first overview and my contribution to the discussion.
3 Likes
Just for clarity, when you use “Add Equipment to Model” to create Items, you get to choose the location and enter the name and label of the Equipment. All of the Item names, by default, become <EquipmentName>_<ChannelName>. The label of the Items by default are the Channel Labels.
And while the Item name cannot be easily changed short of deleting and recreating, there are fewer and fewer cases in the UI where the name of the Item even matters. It’s the label that is the primary thing shown, not the Item name. And the label can be changed at any time. The only times you need the Item name is when you are directly referencing an Item in a script/rule or a UI widget when editing the YAML. In all other cases you have a selection box where you will primarily see the Item’s label.
One place this falls down though is if you have a bunch of the same device and you do not add something to the Item labels to make it clear which device goes with which Item you run into a case where you have a bunch of “Mode” labeled Items which are hard to tell apart in certain places in the UI.
I’m not saying this is the ultimate best way to do things, but it’s not as bad as it’s made out to be.
I’m not saying this is you @iLion, but in general I get frustrated when people who do not use the UI complain about how hard something is to do in the UI because they gave it half a glance a couple years ago. If working in the UI were anything as bad as it’s made out to be, I certainly wouldn’t be using it exclusively. I’m patient but not that patient.
For completeness, here is the flow, with pictures, to go from Thing discovery to UI representations in the Overview tabs.
I’m assuming at least the Locations in the Semantic Model are already created.
-
Navigate to Settings → Things → + → binding → Scan (note some bindings will scan periodically and discover stuff on their own).
-
Pick one or add them all at once. You can change the label for the Thing now.
-
There is room for improvement between the previous step and this step. Now we’ve returned to the Things page. Scroll down or search for your new Thing and select it. Click on the Channels Tab and click “Add Equipment to Model”. (Improvements: automatically offer to add equipment to model, automatically select the Thing after adding it, default to the Channels tab as the default).
-
Select the Location. Enter a meaningful name for the Equipment. Select the Channels desired or select them all. Modify the Items properties as desired. Note that the semantic tags can impact the widgets that will appear by default for that Item. Often a reasonable default is chosen.
If clicking and selecting awkward for you, there is expert mode:
Once all is as you like it click add and you’re Items are created. You can now find them on the Overview tabs.
I kind of randomly selected some semantic tags without paying attention. I should have chosen a point tag of Switch for the Mute Item. Except for the Thing UID and the Item Names, everything shown above can be changed after this step.
At this point we have a mostly reasonable config with minimal work. Can this be improved? Absolutely! But this is the base that we should be working from. And that’s why I posted this. How do we automate more of this choosing reasonable defaults for things like Item names and labels? How do we make these steps more obvious?
And I’ll close with the following observation. KNX, Generic MQTT, HTTP, Exec, Serial, and Modbus (perhaps more) are not typical add-ons. These are low level bindings that require significant amounts of manual configuration and outside knowledge. While experiences with these are valid and should be addressed, saying “openHAB is too hard because it’s hard to configure a Generic MQTT Thing” kind of ignores all the great things that happen with technologies where devices can be auto-discovered, sometimes even without manually configuring a Bridge Thing.
If you want to compare between OH and HA, make it apples to apples. Show the same technology in both.
2 Likes
I wonder why it’s so difficult to implement this in OH. If this is not possible, I like @JustinG idea, could be a feasible alternative.
This, plus integration with Z2M and esphome, are key reasons for HA’s success.
Starting with [RfC] Central UPnP/mDNS thing discovery for suggesting add-ons to install · Issue #2645 · openhab/openhab-core · GitHub would be a good place to see a few of the challenges. But ultimately it’s the same as every other answer in this thread. Someone needs to volunteer to implement it. It wouldn’t be a trivial change.
This would just require someone to add support to the MQTT binding for auto discovery. Again, volunteers are encouraged and welcome.
The last I looked at it ESPHome was using the HomeAssistant standard which can be automatically discovered by OH.
Hi Rich,
it is not good that you are frustrated, because you are of the ones I mentioned with my comment to the nice and helpfully openHAB community. I tried this, what you explained in my Lab some minutes ago, you are right in some way. My Lab was a fresh installation with Ubuntu 22.04 LTS and openHAB 4 I setup to learn about the UI. My “overview” page is not edited because I did not understand how to get these information, I want to see. It is very complex for me personally. And I did not want to destroy my produktiv system.
As one part, I generated a KNX device and there is no Inbox for KNX devices, so I had to do it over the UI manually.
But now I tested it with the shelly binding. First a saw that there is no search in the Binding section. Perhaps I need glasses, but it need time to find the binding in the whole list (Home Assistant offers for integrations a search). After this, nothing happened. I found the scan button and the devices are listed. I clicked on the first one to setup up. The suggested name was the Type with MAC and the IP. I only select “ok” like most users would do and got the name. Then I clicked to the channel section and select to add all channels. Next problem was, that I had to click all checkboxes to do this. Then the resulted names are like you described. In comparison to Home Assistant, there you only have to accept the found Shellys, give them a room and you are ready. But as a look into the future, some month ago I changed my whole network, new IP ranges, and the the names from the UI would not match any more.
To avoid any misunderstanding, for me openHAB is rock solid. It is my working horse. But I also wrote some months ago in a discussion, the reason i more prefer the file based method to get on an easy way fast a clean system. For example, when Ubuntu 24.04 LTS will arrive, I will setup a new VM, setup hopefully openHAB 4.2, copy my files to the folders and start a fresh system. All the things over UI will not help, when using a backup and a database have perhaps a bug, then the bug will restored. But this is not the topic of this discussion, why openHAB have not the marketing, it perhaps needs.
Please do not forget that @seime is working on a great new esphome binding which for me is working brilliantly!!! Auto discovery works and all. It’s not fully compatible with everything but a lot is already working really really well 
Edit: i need to find a way to make multiple quotes from different people in the same comment on a smartphone…
@iLion
In comparison to Home Assistant, there you only have to accept the found Shellys, give them a room and you are ready.
Indeed, I would also like to see this topic hit squarely in the head. For example: we are sorely missing in the helping wizard department, yes. But in large part this is because openHAB is more “manual” than the alternatives. I’d like to see a more “auto” approach being pursued, for example, if you add a new Shelly thing, auto create all items inside of the channels with their auto generated names immediately. If the items are auto created, then users don’t even need to wonder “what is an item”.
1 Like
I’m on the snapshots and this recently changed. In 3.0+ there’s Settings → Bindings.
There was a search icon at the bottom right.
In 4.1 snapshots there is a new menu “Add-on Store” right under “Settings”. Search is still at the bottom right.
An issue was recently filed (I think) to make the search more obvious.
I think how this works depends on the binding implementation. Some automatically start scanning (Chromecast), others wait to manually be told to scan (Network), others require manual configuration of a bridge Thing (anythign with an account or serial connection to a device) and still others require 100% manual configuration (KNX et. al).
It’s hard to enforce consistency because do you really want a Thing for everything that responds to a ping added to your Things Inbox automatically after installing the Network binding?
My understanding is Shelly is a bit weird in this regard. Most bindings would simply add the device with all the Channels already defined and ready to go. Shelly does some weird sort of Channel discovery at runtime which doesn’t always work.
But of all the names in OH, the Thing name matters probably the least.
That’s on the binding I think. Usually there should be a “select all” option. Maybe there isn’t a way for a binding to request that in the config. But again, the fact that you have to add Channels like this at all in the first place is highly unusual and specific to the Shelly binding as far as I’m aware.
OH typically has more steps, yes, but not as many. As I showed above you accept the found Things and create the Equipment as shown above. This whole selecting of Channels is frankly weird.
You have some choices here:
- leave them, who cares? The IP addresses aren’t used in the Item labels and that’s what gets shown almost everywhere anyway.
- when you create the Item/Thing change them to something meaningful (I always recommend this regardless)
- delete the Thing and recreate it with a new ID; unfortunately you’ll have to recreate the Links to the Items since the Thing ID changed
- manually edit the JSONDB files to patch in a new ID; do not do while OH is running.
Agreed it’s off topic, but I don’t understand. Perhaps you have a bug in your text configs and those get copied over too. 
A backup and restore is going to work with what it’s given. Managed configs are stored in files too, just different files.
After you start editing tap the down arrow to collapse the edit pane, highlight the text you want to quote and the “Quote” button will pop up. That quote will be added where ever the cursor is in the edit pane.
Sounds great! So great we tried it! OH 2.x had this feature. It was a freaking nightmare. I’m not saying this cannot work, but it is not just as easy as that.
1 Like
Thanks Lars, I asked for folks who knew HA because I’ve never really tried it. Your description was just what I was looking for, understand the differences.
on another note…
and is frankly… the stupidest and its metaphoric meaning the most incomprehensible for beginners
1 Like
+1. I know that Device is probably not ideal, but it’s a whole lot more understandable than Thing.
And Binding → Device Driver
I got it, way back when, but most do not. And no… I don’t have a better idea then Thing
Add-On and binding is my pet peeve. The two names for the same thing are freely interchanged on this forum and in the documentation but noobs don’t know that!
I think we should all endeavor to say Add-On when we mean binding. Add-On means something that is (sort of) meaningful. Binding sounds like something that holds your under britches up 
Binding is a subset of add-on
There’s Binding Addon, Automation Addon, Transformation addon, UI Addon, etc. So Binding and Addon aren’t equivalent.
Maybe
Add-ons → Extensions (like the term they use in Browsers)
Binding → Adapter ?
I said the same thing to a few of my colleagues too; slight the other way around: I use OH and its the best thing since, sliced bread; highly configurable and sh!toads of integrations, which you need, if you do not want to have umpteen apps on your phone, and when you put smarts into the system. You want to look at HA which might be simpler to use, but is arguably less powerful. I often get the reply: I never heard of OH, but HA. It s the latter crowd that goes for HA. Often telling me later that they had a look at OH but it seems complicated.
This is the biggest bug-bear for me too.
Naming is actually crucial when you want to use group rules and split on underscore to identify and modify other items by building names in rules.
I am in the process of buying 60 Zigbee RGBCW down lights for the house I am building. One can imagine, that a ‘standard’ naming convention will help immensely.
sometimes I start with one device; then add another; then realise I should have named them differently. When I migrated from OH2 to 3, I wanted to rename a bunch of similar devices to a ‘standard’ naming scheme. Guess what? It never happened, because of needing to recreate the things and items. Simply too error-prone.
I later read Rich’s commentary OpenHab Marketing is Lacking - #112 by rlkoshak to the above… Point taken, but still a pain in back side
However, I never saw these two screens: ‘add item from things’ and ‘expert mode’ … oops.
1 Like
I rename heaps on my OH. I didn’t want to recreate it all.
1 Like
hmmm… good point, far enough
still confuses folks
ok, Greg… I didn’t watch your video yet… but, I did read the thread you linked to and thought it had AWESOME all over it
If your videos are anywhere as good, you’re the next francis ford coppola