Hallo Peter,
I am glad to hear you want to continue with that binding! Your stats look promising. I have similar ones as well, thanks to you great work on the binding! What I noticed is that the fan speeds are definitely wrong, but that also applies to my heatpump, so nothing serious, but should be fixed by us at some point.
I think you have found my https://github.com/t2000/openhab2-addons/tree/tecalorBindingWIP branch, right?
I called it WIP because the only stuff I did so far is to make it compile again in the new bnd workspace and I refactored the class names a little to be conform to Java and openHAB standards.
Also I hacked into the binding a thing type for my heatpump a Tecalor 5.5 which has integrated ventilation. Thus as a quick hack I only added test channels to control the ventilation stage for day and night. I basically took the commands from the FHEM integration and hardcoded the values
Regarding the Typeresolver that you were using: I think there is unfortunately a general misunderstanding when it comes to Channelgroups and how they are used by you in the binding. As far as I remember I replaced it with the ChannelTypeRegistry to get the binding working again.
However, this is NOT a solution for a binding that will eventually be integrated into the official openHAB code. The reason is that a binding should NEVER need to query any registries (be it ThingTyperegistry, ThingRegistry, ChannelTypeRegistry, ItemRegistry, or whatever). It should not even know that these exist.
When I said there is a “general misunderstanding” regarding the ChannelGroups I mean this: ChannelGroups in openHAB are designed to group some channels together is order to re-use them in different ThingType definitions. Their existence is just for this one purpose, nothing more. it should take the effort from a binding developer to duplicate channel definitions on each thingtype where the channels occur.
At runtime, i.e. when the openHAB system is running, the Thing object has Channels ONLY. The groups are GONE and there is no way to obtain the information if a Channel was part of a group or not. Basically what happens is that all Channels are glued flat to the Thing. And this is the reason why the concept on how the ThingTypes are defined currently has to be optimized.
I was busy in the past days to finish my binding for the Nova fine dust sensor (https://github.com/openhab/openhab-addons/pull/7528) and once this is running fine in my docker container alongside the heatpump which also creates a ttyUSB device, I can continue to contribute to the stiebelheatpump binding. I really would like to see it as part of the official openHAB distribution.
So I guess the next steps are to think about a concept of getting rid of the “channelgroup misuse” and then take it from there.
If you like we can also have a (voice) chat sometime to discuss things, maybe also in german as I see you are from Germany too. Just let me know if a discussion/brainstorming would be of any help and we can arrange soemthing.
Regards,
Stefan
PS: The “t2000” is actually my private account, whereas “triller-telekom” is my business one, just to solve the confusion why I post with 2 nicknames