gen.Server HandleInfo run in different goroutine

[xiaodoudou12]

https://github.com/ergo-services/ergo/blob/master/gen/server.go +430

func (sp *ServerProcess) waitCallbackOrDeferr(message interface{}) {
	if sp.waitReply != nil {
		// already waiting for reply. deferr this message
		deferred := ProcessMailboxMessage{
			Message: message,
		}
		select {
		case sp.deferred <- deferred:
			// do nothing
		default:
			lib.Warning("deferred mailbox of %s[%q] is full. dropped message %v",
				sp.Self(), sp.Name(), message)
		}

		return
	}

	switch m := message.(type) {
	case handleCallMessage:
		go func() {
			sp.counter++
			sp.handleCall(m)
			sp.callbackWaitReply <- nil
		}()
	case handleCastMessage:
		go func() {
			sp.counter++
			sp.handleCast(m)
			sp.callbackWaitReply <- nil
		}()
	case handleInfoMessage:
		go func() {
			sp.counter++
			sp.handleInfo(m)
			sp.callbackWaitReply <- nil
		}()
	case ProcessDirectMessage:
		go func() {
			sp.counter++
			sp.handleDirect(m)
			sp.callbackWaitReply <- nil
		}()
	case nil:
		// it was called just to read the channel sp.callbackWaitReply

	default:
		lib.Warning("unknown message type in waitCallbackOrDeferr: %#v", message)
		return
	}

I don't understand why start a new goroutine when ServerProcess receive CastMessage or InfoMessage.
I think ServerProcess is like a Process in erlang,gen.Server like a gen_server in erlang,as a actor, HandleInfo and HandleCall should run in the same process.

[halturin]

Thanks for your question. We should be able to handle incoming messages in case of making sync requests during callback handling (e.g., do process.Call(to, msg) within HandleCast callback). Basically, any sync request is just a pair of async messages with some internal lock.