Jul 1, 2019

Stateful Programming

#programming

I’ve heard people say that β€œstate” is the cause of many bugs. This weekend, I came across a good example of describing a stateful program vs a stateless program that shows some of the tradeoffs.

Scenario

You get on a bus from Home and are looking to get off the bus at Baker Street. You can figure out when to get off the bus in a couple of ways:

  1. Figure out the number of stops to Baker Street and get off after the bus has stopped that many times.
  2. Remember the name of your stop and check every stop to see if it's that one.

If you use the first method, you'll have to remember how many stops have passed. This is state: remembering information for use later. If you use the second method, you'll have to pay attention every time the bus stops to see if your stop has been reached.

Code

Let's see if we can write this in code. We know the bus route, and we know where we are getting on and getting off:

const ROUTE = ["Before Home", "Home", "Stop1", "Stop3", "Baker Street"];
const ENDING_STOP = "Baker Street";
const STARTING_STOP = "Home";

And we can listen for a stop event to decide whether or not we want to get off like this:

window.addEventListener("stop", (stopName) => {
    // handle the event
});

To handle this using the stateful method, we can do something like this:

// calculate the number of stops
const remainingStops = ROUTE[ENDING_STOP] - ROUTE[STARTING_STOP];

function(remainingStops) {
    this.remainingStops = remainingStops;
    window.addEventListener('stop', () => {
        if (this.remainingStops === 1) {
            alert('get off the bus!');
        }
        this.remainingStops--;
    });
};

And to handle this in a stateless way, we can:

window.addEventListener("stop", (stopName) => {
    if (stopName === ENDING_STOP) {
        alert("get off the bus!");
    }
});

You can see in the stateful example, that the stopName at each stop does not matter and is ignored by the event handler; the handler only cares about the number of stops remaining. The stateless handler, however, requires a comparison to the ENDING_STOP at every stop.

Tradeoffs and Insights

There are many ways for either of these programs to fail and for you to miss your stop. But some of those ways are specific to the method you use.

For example, if you fall asleep on the bus ride and wake up when the bus is already past Baker Street, you will have missed your stop regardless of the method you chose. However, if you fall asleep and wake up before Baker Street, your count of "remaining stops" is now invalid, but you can still compare each stop and reach Baker Street safely.

If you calculate the number of remaining stops incorrectly, it's completely unpredictable what will happen, since any number of other things could make your counter correct or incorrect. With the stateless method, it's much more likely that you get off at the right spot.

On the other hand, the stateless method means that every time the bus stops, you have to look outside the window to see where you are and if it's your stop. If you forget to look up or can't look up because your coffee just spilled or you're distracted, you could easily miss your stop. The stateful method in this case would save you, since you could know that your counter is at 0 without looking up.

By now, it should be fairly obvious to most people that counting down the number of stops and looking out the window are not mutually exclusive ways of determining whether you've reached your stop. You may use a combination of the two methods, such as looking outside every third stop to get your bearings and revalidating your counter. You may use heuristics such as knowing the approximate length of the ride to determine how long you should count for, and when you should look at where you are. You might also optimize for different things: maybe you want to read a book and it doesn't matter if you miss your stop, or maybe you really can't miss your stop today and need to be vigilant.

Another interesting insight is that whichever method you choose, it's not the only place to find the answer you're looking for. You could ask the bus driver or a fellow passenger how many stops there are left. You could use an app to send you a push notification when your stop is approaching. You could even just get off at the wrong stop and walk from there or take the next bus!

Conclusion

Our brain not only handles and processes a huge number of scenarios when doing something as simple as taking the bus, it is also able to use a number of different approaches to solving the same problem and merge the results of them together. All of these approaches together is what makes the "program" successful.

But when I write code, I rarely use more than one approach to solving the problem (although I might explore more than one solution), and rarely handle more than a handful of error scenarios. This isn't necessarily bad, to be fair. One reason for this is that the amount of data available to our program is much more limited than the amount available to our brain. Another reason is that we rely on contracts from the things external to our program. For example, we may expect that a bus will only ever stop at one of the stops in the known route. If it stops anywhere else, we'll get off at the wrong stop, but it won't be "our fault"--we'll just complain to the people who own and operate the bus.

Constraints or optimizations like these are easy to understand at the time of writing a program, but it's important to understand that they may be different every time the program is run. Choosing to write a program in a stateful or stateless way may depend on the constraints and desired optimizations at the time of writing, but over time, as both of these change, unexpected behavior (i.e. bugs), can easily appear. As this bus ride example illustrates, the most resilient and optimal way of solving a a problem is probably a complex combination of a stateful and stateless program. Until we can achieve that, however, the important thing to remember is that either method on its own will have its tradeoffs, and the job of the programmer is to figure out which tradeoff to make at any given time.

If you like this post, please share it on Twitter and/or subscribe to my RSS feed. Or don't, that's also ok.