10 May
2024
10 May
'24
8:53 a.m.
пт, 10 мая 2024 г. в 08:25, Arno Schoedl via Boost
Absolutely agreed. The current coroutines have their place but that we effectively need to keep two copies of every generic algorithm around, one plain and one coroutinified, just in case somewhere deep inside the callstack a caller-supplied lambda wants to yield is very unsatisfactory.
I was reliably informed in this paper (https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/p0171r0.html#island...) that the problem you are describing does not in fact exist.