Scheduler

The scheduler class — available via require — can be used to easily set up coroutine-based cooperative multitasking.

__construct

The constructor takes no arguments.

add

The add method can be used to turn a function into a coroutine. The coroutine is resumed once and then added to the scheduler.

addloop

The addloop method is a wrapper for the add method that keeps invoking the given function until it returns false.

run

The run method activates the scheduler. The scheduler resumes all coroutines and then calls its yieldfunc until either all coroutines are finished, or coroutine throws an error — in which case it is rethrown.

local scheduler = new (require"pluto:scheduler") ()

local loops = 0
scheduler:addloop(function()
    ++loops
    print(loops)
    return loops != 5
end)
scheduler:run()
--> 1
--> 2
--> 3
--> 4
--> 5

scheduler:add(function()
    print("3...")
    coroutine.sleep(1000)
    print("2...")
    coroutine.sleep(1000)
    print("1...")
    coroutine.sleep(1000)
    print("Boom!")
end)
scheduler:run()
--> 3...
--> 2...
--> 1...
--> Boom!

yieldfunc

yieldfunc is a property on the scheduler that can be used to customize how the scheduler itself yields. By default, this is a function calling os.sleep(1).

local scheduler = new (require"pluto:scheduler") ()

-- Yield to OS every second instead of every millisecond.
scheduler.yieldfunc = || -> os.sleep(1000)

local loops = 0
scheduler:addloop(function()
    ++loops
    print(loops)
    return loops != 5
end)
scheduler:run()
--> 1
--> 2
--> 3
--> 4
--> 5

errorfunc

errorfunc is a property on the scheduler that can optionally be set to customize how the scheduler reports errors in its coroutines. By default, this is nil, in which case any coroutine errors are rethrown (via coroutine.xresume).

local { scheduler } = require "*"

local sched = new scheduler()
sched.errorfunc = |err| -> do print(debug.traceback(err)) end
sched:add(function()
    error("this is urgent: fatal error")
end)