Future destructors normally just destroy the future’s data members, execept for the final future referring to a shared state for a non-deferred task launched via std::async, which blocks until the task completes.

Join-on-destruction can lead to difficult-to-debug performance anomalies; while detach-on-destruction can lead to difficult-to-debug undefined behavior.

The flexibility that default policy for std::async permits both async and sync task execution leads to uncertainty when accessing thread_locals, implies that the task may never execute, and affects program logic for timeout-based wait calls.

Thread-based programming calls for manual management of thread exhaustion, oversubscription, load balancing, and adaptation to new platforms, while task-based programming via std::async with the default launch policy handles most of the issues for us.

Lambdas are more readable, more expressive, and may be more efficient than using std::bind.