How do I simulate a buffered peripheral device with SwingWorker?

Question

I'm using this exercise as a pedagogical tool to help me burn in some Java GUI programming concepts. What I'm looking for is a general understanding, rather than a detailed solution to one specific problem. I expect that coding this "right" will teach me a lot about how to approach future multi-threaded problems. If this is too general for this forum, possibly it belongs in Programmers?

I'm simulating a card reader. It has a GUI, allowing us to load cards into the hopper and press Start and so forth, but its main "client" is the CPU, running on a separate thread and requesting cards.

The card reader maintains a single buffer. If a card request comes in and the buffer is empty, the card reader must read a card from the hopper (which takes 1/4 of a second, this being 1962). After the card has been read into the buffer, the card reader sends the buffer to the CPU, and immediately initiates another buffer-loading operation, in advance of the next request.

If not only the buffer is empty but there are no cards in the hopper, then we must wait until the operator has placed a deck in the hopper and pressed Start (which always initiates a buffer-load operation).

In my implementation, card requests are sent to the card reader in the form of invokeLater() Runnables being queued on the EDT. At myRunnable.run() time, either a buffer will be available (in which case we can send it to the CPU and kick off another buffer-load operation), or the buffer will be empty. What if it's empty?

Two possibilities: (a) there's already a buffer-load operation in flight, or (b) the card hopper is empty (or hasn't been started). In either case, it's not acceptable to keep the EDT waiting. The work (and the waiting) must be done on a background thread.

For the sake of simplicity, I tried spawning a SwingWorker in response to every card request, regardless of the status of the buffer. The pseudocode was:

SwingWorker worker = new SwingWorker<Void, Void>() {
    public Void doInBackground() throws Exception {
        if (buffer.isEmpty()) {
            /*
             * fill() takes 1/4 second (simulated by Thread.sleep)
             * or possibly minutes if we need to have another 
             * card deck mounted by operator.
             */
            buffer.fill();
        }
        Card card = buffer.get(); // empties buffer
        /*
         * Send card to CPU
         */
        CPU.sendMessage(card); // <== (A) put card in msg queue
        /* 
         * Possible race window here!!
         */
        buffer.fill(); //         <== (B) pre-fetch next card
        return null;
    }
};
worker.execute();

This produced some odd timing effects - due, I suspect, to a buffer.fill() race that could occur as follows: if, between (A) and (B), the CPU received the card, sent a request for another one, and had another SwingWorker thread spawned on its behalf, then there might be two threads simultaneously trying to fill the buffer. [Removing the pre-fetch call at (B) solved that.]

So I think spawning a SwingWorker thread for every read is wrong. The buffering and sending of cards must be serialized in a single thread. That thread must attempt to pre-fetch a buffer, and must be able to wait and resume if we run out of cards and have to wait for more to be placed in the hopper. I suspect that SwingWorker has what is required to be a long-running background thread to handle this, but I'm not quite there yet.

Assuming a SwingWorker thread is the way to go, how might I implement this, eliminating delay on the EDT, allowing the thread to block awaiting a hopper refill, and handling the uncertainty of whether buffer-filling completes before or after another card request arrives?

---

EDIT: I got an answer from another thread and will recap it here:

Instead of using a SwingWorker thread, it was recommended I create an ExecutorService newSingleThreadExecutor() once, at the beginning, and have the GUI enqueue lengthy methods on it using execute(Runnable foo), as follows (this code runs in the EDT):

private ExecutorService executorService;
::
/*
 * In constructor: create the thread
 */
executorService = Executors.newSingleThreadExecutor();
::
/*
 * When EDT receives a request for a card it calls readCard(),
 * which queues the work out to the *single* thread.
 */
public void readCard() throws Exception {
    executorService.execute(new Runnable() {
        public void run() {
            if (buffer.isEmpty()) {
                /*
                 * fill() takes 1/4 second (simulated by Thread.sleep)
                 * or possibly minutes if we need to have another 
                 * card deck mounted by operator.
                 */
                buffer.fill();
            }
            Card card = buffer.get(); // empties buffer
            /*
             * Send card to CPU
             */
            CPU.sendMessage(card); // <== (A) put card in msg queue
            /* 
             * No race!  Next request will run on same thread, after us.
             */
            buffer.fill(); //         <== (B) pre-fetch next card
            return;
        }
    });
}

The main difference between this and SwingWorker is that this ensures there's only one worker thread.

Answer 1

It may help to know that SwingWorker uses an ExecutorService internally; it adds the interim EDT processing mechanism for convenience. As long as you update your GUI on the EDT and synchronize access to any shared data, the latter is equivalent to the former.

Assuming you are using the Model–View–Controller pattern, suggested here, your model is the operation of a cpu. Although it may be a different class, I can't see any reason to model the card reader on a different thread. Instead, let the processor model have a card reader model that does the waiting on a java.util.Timer thread, updating the model as the timer fires. Let the updated model notify the view in the normal course of posting events to the EDT. Let the controller cancel and schedule the card reader model in response to view gestures.