I've seen create_task used in a couple ways:
void Bob()
{
create_task() { /* do stuff */ }.then([](){ /* do more stuff */ });
}
and
task<void> Bob()
{
return create_task() { /* do stuff */ }.then([](){ /* do more stuff */ });
}
Why bother returning the task (in the second example) when asynchronous behavior can be achieved with either approach?
Clarification: I'm not asking about the void return type specifically. It could be an int, object, or something else.
void/task<void> is a special case here, because you can magic a void from nowhere. You couldn't do the same with a int, std::string or similar.
void Bob()
{
create_task() { /* do stuff */ }.then([](){ /* do more stuff */ });
}
After this has returned /* do stuff */ and /* do more stuff */ have started, and any handle on their progress is discarded.
task<void> Bob()
{
return create_task() { /* do stuff */ }.then([](){ /* do more stuff */ });
}
After this has returned /* do stuff */ and /* do more stuff */ have started, and you have a handle to wait for them to finish.
int Alice()
{
return create_task() { /* do stuff */ }.then([](){ /* do more stuff */ return 42; }).get();
}
After this has returned /* do stuff */ and /* do more stuff */ have finished, with a final result available.
task<int> Alice()
{
return create_task() { /* do stuff */ }.then([](){ /* do more stuff */ return 42; });
}
After this has returned /* do stuff */ and /* do more stuff */ have started, and you have a handle to wait for them to finish, and get the result.