Tasks, channels, and producer lifecycles

The idea

Tasks allow computations to pause and resume while waiting for other work. A Channel carries values between producers and consumers with explicit capacity and element type. A producer can wait when the channel has no room, and a consumer can wait when no value is available. A zero-capacity channel synchronizes a handoff rather than storing a backlog.

The Channel constructor that runs a producer function binds its lifetime to that task and closes the channel when the task terminates. Iterating the channel consumes values until closure and exhaustion, so completion is part of the protocol. A task is not by itself a promise of simultaneous execution on multiple CPU cores. Use deliberate threading mechanisms where appropriate, and still account for shared state, failure propagation, and shutdown.

ch = Channel{Int}(0) do output
    for n in 1:3
        put!(output, 2 * n)
    end
end
println(sum(ch))

12

The producer sends 2, 4, and 6; its bound task then terminates, closes the channel, and lets the consumer finish.