Should ki force the return value of a thread to WHNF?

[mitchellwrosen]

I wonder if ki should force a thread's result to WHNF before returning.

This would be inconsistent with async, and probably other threading libraries, so by the principle of least surprise, maybe we shouldn't do this.

However, it might be nice to assist users in performing (some) computation on the dedicated thread they explicitly spawn, rather than let them accidentally conclude a background thread with some expensive thunk that's computed by the first awaiter who peeks at it.

[ocharles]

One option might be to add fork', using the convention that ' means strict