One of my personal litmus tests for the executor-future interaction—and I suspect a reasonable litmus test for other committee members as well—is whether inline_executor and executor_future_t<inline_executor> can be written so that they compile away (e.g., so that a program that deterministically does nothing compiles into xor %eax, %eax; ret when full optimizations are enabled). I think a demo of this sort would go a long way towards convincing people that our design is conducive to the zero-overhead abstraction ideal that we aim for in C++.
One of my personal litmus tests for the executor-future interaction—and I suspect a reasonable litmus test for other committee members as well—is whether
inline_executorandexecutor_future_t<inline_executor>can be written so that they compile away (e.g., so that a program that deterministically does nothing compiles intoxor %eax, %eax; retwhen full optimizations are enabled). I think a demo of this sort would go a long way towards convincing people that our design is conducive to the zero-overhead abstraction ideal that we aim for in C++.