You need to update this for recent semtech package changes. I.e. just apply the same change as here.

I see that this is very similar to the lorawan example. My thoughts are however that this example could be designed more towards presenting a clean, extensible interface to the user. Personally I like the structure of e.g. the gcoap and dtls-echo examples - they put a simple interface to the program in main.c. Then, if people want to change other aspects of how the program responds, they can go into the other .c files.

I would suggest finding a program structure so that in main.c they can do something like:

• add measurements to be sent in a declarative manner in a simple macro or function (for Sensebox this could also add the drivers at compile time, not sure whether this would require a separate macro/function or whether it could be done automatically). They could perhaps specify that the application sends dummy data (as above), or direct from sensor, or from a custom function
• change the frequency of data sending

Then in another file (loramac_send.c?)

• asyncronous data sending stuff incl thread definition perhaps
• arrays storing keys

Basically the philosophy I'm suggesting is to present a really simple API to a user in main.c which allows them to quickly, easily and flexibly define what data is going to be sent. As easy as possible. E.g. as little as possible above just plugging together their sensebox and loading this application.