Question

I find myself again and again faced with a similar problem: there’s some piece of code that processes data as it arrives from the user/network/produces of some sort. For efficiency reasons, I don’t want to call flush() or commit() on every piece of data that I receive, but only occasionally.

I usually come up with code like this:

class Processor {
    private final static MAX_SAVE_PERIOD = 60000;
    private final static MIN_SAVE_PERIOD = 20000;

    private final static int MAX_BUFFER = 10000;
    Arraylist<Data> dataBuffer = new Arraylist<Data>();

    private long lastSave = 0;

    public Saver() {
        new Timer().schedule(new TimerTask() {
            periodicSave();
        }, MAX_SAVE_PERIOD, MAX_SAVE_PERIOD);
        Runtime.getRuntime().addShutdownHook(new Thread(new Runnable() {
            public void run() {
                periodicSave();
            }
        }));
    }

    public synchronized void processData(Data data) {
        dataBuffer.add(data);
        if(dataBuffer.size() >= MAX_BUFFER) {
            saveData();
        }
    }

    private synchronzied void periodicSave() {
        if(!dataBuffer.isEmpty()) {
            saveData();
        }
    }

    private void saveData() {
        if (System.currentTimeMillis() - lastSave < MIN_SAVE_PERIOD) return;

        ...        

        lastSave = System.currentTimeMillis();
    }
}

I get the distinct feeling that I’m reinventing the wheel every time I write this, and what’s more, I keep changing stuff every time I write this kind of code, depending if the various parts make sense in the specific context.

It seems to me that this is a very common pattern, but I don’t remember seeing it named or implemented as a library utility. As long as I have to implement this myself, I keep facing analysis paralysis whenever I reimplement it. Please help me!

UPDATE: After I wrote this, I realized that I haven’t accounted for flushing the buffer before JVM shutdown, so I added a shutdown hook in the constructor. Now I’ve realized that this code will still not work properly if the shutdown happens less than MIN_SAVE_PERIOD milliseconds after the last save, so I should probably refactor saveData. It’s happening again.

Answer 1

Your code already says it: it’s called buffering.