I have replaced the function layer with two separate layers. Key transparency from the default layer to the first layer works as expected. Transparency from keys defined in the first layer to the second doesn’t work as I would expect. The second layer is getting keys from the default layer rather than the one below.
Transparency gets you keys from the active layers below. If you have three layers, with 0 and 2 active, transparent keys on layer 2 will be from layer 0. If you make layer 1 active too, then the transparent keys of layer 2 will be looked for on layer 1. If they are still transparent there, then from layer 0.
Argh. This is the metaphor issue we were talking about here:
I’ll adjust the intro document to be more correct right now. Might make sense to make a comment about this in the Model01-Firmware.ino as well. I am still not convinced that renaming the function names is the best course, but this behavior is going to confuse a lot of smart people, and we should think seriously about how to mitigate it.