如何将 WaitGroup 传递给连续的 function 调用？

Question

I have a function which can be called sequentially or concurrently in separate goroutine.

I want to ensure that function is executed completely before main goroutine finishes, so I am passing *sync.WaitGroup argument to the function. Now, at some places the function is to be called sequentially.

I can pass nil waitGroup to the function like this:

func my_func(wg *sync.WaitGroup){
   if wg != nil{
   defer wg.Done()
   }
   // do the task
}

func main(){
my_func(nil) // sequential call

wg := sync.WaitGroup{}
wg.Add(1)
go my_func(&wg)  // concurrent call
wg.Wait()
}

Is there any better way to achieve this?

Answer 1

Your my_func() should not know / should not care how it is executed (whether in a new goroutine or not). So just for this you should not pass wg . Do not enforce concurrent or non-concurrent use of your API, let users of your package decide how they wish to call it.

If someone wishes to run it concurrently, in a new goroutine, wg can be handled outside of my_func() like this:

wg.Add(1)
go func() {
    defer wg.Done()
    my_func()
}()

This also gives possibility to put further code before / after the function call that will be executed before the wg.Done() call:

wg.Add(1)
go func() {
    defer wg.Done()

    // other code before
    my_func()
    // other code after
}()

Also note that if you have this in many places, you can create a helper function that takes care of the goroutine launching and waitgroup handling:

func launchMyFunc(wg *sync.WaitGroup) {
    go func() {
        defer wg.Done()

        my_func()
    }()
}

You can also create a helper that accepts an arbitrary no-arg no-return function:

func launchFunc(wg *sync.WaitGroup, f func()) {
    go func() {
        defer wg.Done()

        f()
    }()
}

Using the above helpers this is how you could do the same:

wg.Add(1)
launchMyFunc(wg)

// or
wg.Add(1)
launchFunc(wg, my_func)