ZigBee is way too complicated and messed up for hobby. Atmel/Microchip have ZigBee stack that works with GCC, but I think it is only available on request after the acquisition.
But you really don't want to get involved with ZigBee if you don't have to. It would be a huge waste of time.
Yes, I agree Zigbee is not simple to mess with. But I'm an engineer in IT and I'm not scared about software complexity. This project would not be something I could get over the week-end but something it will take me months to complete. I do it mostly for the challenge more than getting things done. I'm an engineer, and I like to over-engineer as a result
.
There are XBee modules which do Zigbee https://www.digikey.com/short/q7jv0b. There is an open source library https://github.com/digidotcom/xbee_ansic_library that facilitates creating a ZCL implementation on your product as well.
Yes I know about XBee solutions. But I don't like this solutions when you rely on some other work and don't know how it work end to end. I do electronics more for a quest of knowledge and master than the pleasure to do things. I like when it end up with my own PCB and case, almost like a professional solution.
If I would like just to do things I would surely use their work.
Why would anyone use the Zigbee radio channel in 2017? There are Zigbee apps protocols, which make sense running over various IP channels, but the actual radio protocol is pretty much dead.
Sorry, I don't understand what you mean. Do you mean using directly the Zigbee radio for proprietary protocol is dead and Application Profiles are the way to go ? If so, I totaly agree with you. Application Profiles seem what will revive Zigbee. And I think it could set it as the home standard for LAN IoT.
I wanted to experiment with Zigbee for its low power but also to make things which would be easily compatible with any Zigbee compliant solution. For the moment, I'm more aimed at an OpenHab raspberry pi solution.