Is there a use case for a changing master node?

[2bndy5]

I was testing the exposed addrList in pyrf24 and hit a segmentation fault when I changed the mesh child node into a master node and explicitly called loadDHCP(). I already can see the problem because if the node's ID is not 0 upon calling mesh.begin(), then the addrList doesn't have any memory allocated.

code to reproduce segmentation fault

```cpp mesh.setNodeID(1); mesh.begin(); mesh.setNodeID(0); mesh.renewAddress(); mesh.loadDHCP(); ```

Is there a practical use case in which the master node of a mesh can change similar to how child nodes change?

This wouldn't be a hard problem to solve, but it would definitely involve increased compile size... One could probably just solve it by recalling mesh.begin() after mesh.setNodeID(0).

[TMRh20]

I don't think there is really a use case for it, plus as you mentioned it would increase memory and program space usage just to support an unlikely scenario.

[2bndy5]

I also had an idea a while back (when porting to pure python) to reserve a msg type that transfers the DHCP list to a new master node. The only case for that would be if you want a continuous network running while phasing in a firmware update.