I don't think the original comment meant long polling (i.e. keeping the connection alive), they meant periodically call the endpoint to check for events.