youāre right, the way weāre requiring engines to be subclasses of CroneEngine essentially makes SC as we use it into a compiled language. and i agree, debugging SC isnāt the best experience and involves looking at a lot of useless / misleading stack traces. (i also find it easier to debug classes than interpreted code, but YMMV.)
on one hand it might be good to extend the system to allow engines to be created in the interpreter. on the other hand, 1) i donāt know what that wwould look like exactly (plz suggest), 2) the class system is really helpful for reliability / consistency / testability, 3) thereās no IDE on the device anyway (and its a long road to make maiden as usable as scide or scel for SC work)
that said,
absolutely. as far as i can tell, all the norns SC classes ājust workā on the laptop. (just install them in your platformās āExtensionsā folder. [+]) no emulation is really required since itās all OSC. i definitely recommend developing engines off the device first. there are some bits and bobs in the repo showing all-SC engine tests.
[+] oh! sorry. there are also the custom faust-generated UGens which need to be compiled and installed, or SC will complain. IIRC the waf scripts to build them work on linux and macOS, but you need to manually install the resulting .so/.scx files. these ugens are gone in new-crone. (baked into the jack client.) [ed: as pointed out below, these arenāt even necessary for engine development and the error messages can be ignored.]
matron also mostly runs on other linuxes but without screen/encoders/buttons. some of the rPi work could be applied to other linux environments (e.g. cairo layer.)