openHAB 2.0 binding

[andreacioni]

@gonzalonal I've implemented the check on startup. On serial gateway with the last version I'm not having any issue, so I would like to know: what hardware are you using? I think that probably is only a timing issue because, for exaple, Arduino UNO reset itself on opening connection and the message for request I_VERSION is sent immediately after connection is opened. I'll suggest you to try adding something like this:

Thread.sleep(3000)

After line 49 on MySensorsSerialConnection.java. Should become:

serialConnection = new NRSerialPort(serialPort, baudRate);
        if (serialConnection.connect()) {
            logger.debug("Successfully connected to serial port.");
            Thread.sleep(3000); 
            mysConReader = new MySensorsSerialReader(serialConnection.getInputStream(), this);
            mysConWriter = new MySensorsSerialWriter(serialConnection.getOutputStream(), this, sendDelay);

            connected = startReaderWriterThread(mysConReader, mysConWriter);
        } else {
            logger.error("Can't connect to serial port. Wrong port?");
        }

I can't do this test because I do not have an appropriate hardware, can you try the suggested code modification?

Thanks!

UPDATE: I've upload a temp build here: https://drive.google.com/file/d/0B11IarucpGdUdFJ2ZjVtZENsb2c/view?usp=sharing if you can't do the modification you can use this. Let me know!

[gonzalonal]

Hi @andreacioni. Thanks for your answer.
I have tested your custom build and it's working flawlessly.

09:10:36.913 [DEBUG] [g.mysensors.protocol.MySensorsReader] - 0;0;3;0;14;Gateway startup complete.
09:10:37.161 [DEBUG] [g.mysensors.protocol.MySensorsWriter] - Sending to MySensors: 0;0;3;0;2;

09:10:37.172 [DEBUG] [g.mysensors.protocol.MySensorsReader] - 0;0;3;0;2;1.5.3
09:10:37.177 [DEBUG] [s.internal.MySensorsBridgeConnection] - Good,Gateway is up and running! (Ver:1.5.3)

Let me tell you about my hardware configuration. I am running OH2 Beta2 in a Raspberry Pi 2 Model B with an Arduino Nano V3 working as Serial Gateway directly connected to one of Rpi USB ports.

Please, let me know if you need me to some tests or maybe is just my config.
Thanks, regards.

[andreacioni]

@gonzalonal I've done all of my test on a custom ATMEGA328p shield for BeagleBone Black, running OH2 (beta) so the environment is the same but I've not the reset circuit that gives the issue above. So now I'll push this little fix on the github repo, I think there's nothing to to in this sense. Thank for your time :) !

If you find some other strange behaviour also remember that github repo has it's own issue tracker available here: https://github.com/tobof/openhab2-addons/issues, it works very well and is the best place to keep track of every software-releated bug :+1:

Greetings!

[TimO]

There is a new build of the jar available!

Changes:

• Added delay before checking the SerialGW (@andreacioni )
• Requests are answered by the binding (The binding returns "0" if no value was set before)

https://github.com/tobof/openhab2-addons/issues/8

[kolaf]

Are any of you having trouble with this binding together with other bindings that use serial ports? It looks as if the different bindings try to use the same serial port lock, so I get a conflict where things fail til initialize. Think it had something to do with the serial port library...

[TimO]

@kolaf Is it zwave again? Last time I tried to fix the conflict I needed to change the zwave binding.

[kolaf]

It is both zwave and rfxcom. I played about it in the oh2 community some time ago, let me see if I can find a link.

[kolaf]

It is my post: https://community.openhab.org/t/rfxcom-2-0-binding/4427/28?u=frankose

[gonzalonal]

Hello @TimO .

Just tested the last build of the binding. Works great.
I see you manage to answer the nodes requests without having to trigger a rule to respond.

I though that wasn't possible due to the limitation of the API. Did they manage to solve this from the OH side?

Does the binding itself keeps track of the items values, or does it querys OH (or its persistence service) for getting the right value?

Thanks!
Gonzalo

[TimO]

Hi @gonzalonal !

I'm glad it works, thanks for testing!
Sadly the limitation of the API is still there, but for the implementation of the requestAck feature I needed to implement a memory and now reused this memory to answer requests.

There is a really big BUT coming with this implementation: we currently can't use persistence out of the box. For this a rule is needed. After startup of OH we need to use "sendCommand(ITEMNAME, VALUE)", where VALUE Is the value stored in persistance. With that the memory of the binding is filled initially.
Hopefully I will find an easier way in the future .

[gonzalonal]

I get what you mean @TimO .

Unfortunatelly, using the rule method you suggest would create a lot of traffic (depending on how many sensors you have) at boot up time. Maybe there are some sensors that wouldn't behave the intended way if receiving the "last value" repeteadly.

An example. Lets supose I have an automatic gate, that everytime it receives a command, wether its a "1" or "0" it toggles its moving direction. Like the most common automatic gates remotes.

Gate is closed.

Press button > Gate opens
Press button > (if gate moving) Gate stops
Press button > (if gate open) Gate closes

Now, suppose last command send to the gate made it close. Suddenly, OH restarts itself because of power failure or whatever. It runs the rule to load the memory of the binding, while at the same time it sends to the sensors the last status taken from the persistance service. This would make the gate to open without your notice.

This may create dangerous escenarios.
I hope you can understand what I mean.

Maybe it would be a better idea to handle requests with rules in the way I suggested in the github issue, that is the same way as the OH-1 mysensors serial binding works.

Let me know if I can help in any way.
Regards.

Gonzalo.

[TimO]

@gonzalonal Good point! :worried:

If a request (or a message in general) is received there is no "commandReceived" in OH, so you can't trigger a rule based on a message received. So doing it with a rule is not possible.

A workaround would be to use "postUpdate()" to update the value after restart. This isn't implemented yet, but is no big deal. This would be a workaround for the workaround, because we still need the rule. :confused:

In the OH1 binding the request is answered with the state of the item, as I would like to to do too.
Line 277: https://github.com/bloft/openhab/blob/master/bundles/binding/org.openhab.binding.mysensors/src/main/java/org/openhab/binding/mysensors/internal/MySensorsBinding.java

[gonzalonal]

Hi @TimO . Good we got in the same page.

I think this topic needs some more thinking.

I was reading OH wiki and found this about receivedCommand.

 Implicit Variables inside the Execution Block

Besides the implicitly available variables for items and commands/states (see the script documentation), rules can have additional pre-defined variables, depending on their triggers:


* Every rule that has at least one command event trigger, will have the variable 
receivedCommand available, which can be used inside the execution block.

* Every rule that has at least one status change event trigger, will have the variable
previousState available, which can be used inside the execution block.

https://github.com/openhab/openhab/wiki/Rules

Maybe this is not available anymore, or at least on OH2.

For now, I think it would be a better approach to use postUpdate at systemStart to populate the binding memory. With this, we'll avoid sending commands to the nodes.

Regards!

[gonzalonal]

Hi again @TimO .

Just an idea.

What do you think about solving this problematic in this way:

We create a group whose members are all those items that later will populate the binding's memory. I have a group named mysqlpersisted where I put all the items whose data will be stored in my SQL DB. So I could use that group. For this example, let's call it "MemoryGroup"

Then, we can create the following rule:

rule "Populate MySensors binding memory"

when 
System started
then

(maybe add a delay to allow persistance services to load)

MemoryGroup?.members.forEach[i |
        postUpdate(i.name, i.previousState(true))
]

Using this method, memory would get populated everytime OH boots up, with data belonging to the persistance service that we use. And we use postUpdate, so no command is sent to the nodes.

With this I think we solve both problems as data in the binding will be concurrent with the persistance service data.

Let me know what you think about this.

Regards!

[TimO]

Hi @gonzalonal !

I was thinking in the same direction and I have played a litte bit with different rules. I had no success in finding a rule that worked. Additionally I don't like the workaround. ;-)

A good solution would be, to trigger a rule, if a request is received, but there too I had no success till now.

I'm still struggling with it ...

[tomkxy]

@TimO Do you think it would be hard to add MQTT support to the binding? I had a brief look to your code and it looks like the protocols are nicely isolated.

[TimO]

I'm currently trying to migrate my production environment from OH1 to OH2. I'm now using two serial gateways.

@kolaf : The problem you described occured to me too. In my case I use two serial gateways. I've added a little method that ensures that a connection to more than two serial ports via nrjavaserial is possible. It works within the OH2 binding. Could you test if it works with the RFXCOM binding you are using? A new jar is available on github. OH2 is currently using an old version of nrserialjava that has a bug. I don't know how to switch the version yet.

@tomkxy : I will integrate MQTT in my OH2 Installation soon and will come back to your request after that.

[undee]

The binding doesn't seem to handle the "requestTime" function yet. Time-aware sensors would be able to handle temporary OpenHAB outages by executing some scheduled actions autonomously, which I could use in the window cover controllers I'm developing.

[kolaf]

@TimO Thanks for the update. Unfortunately it did not seem to change anything for me, rfxcom still fails trying to use the wrong lock file. If it is helpful I can send you the log. Appreciate your input and effort.

[kolaf]

The weird thing is that it seems to work with an older version of OH2 (and an old version of your binding), but not with the more recent beta releases.