IssueTracker

Reactor has historically not supported creating multiple routines and making calls between them. That is, it's not a standalone 'language', it requires a host language (C++) to make the routine calls. That partially changed when support was added for making calls through function pointers, which enabled trampolines. But it still requires some boilerplate code for obtaining the function pointer, and two explicit Call()s.

It might be feasible to roll this into one. When a lambda function produces a Routine, or even a Function, we could automatically first make the call to the lambda, cache the result inline, and make the function pointer call. The Call<>() could also take a caching strategy template parameter in case we want more than just inline caching.