educative.io

Why we do not wait() in release methods?

Courses
,
#1

When I look at this problem, I write a version here:

dWriteLock {

    int readers = 0;
    boolean isWriteLocked = false;

    public synchronized void acquireReadLock() throws InterruptedException {
        while (isWriteLocked) {
            wait();
        }
        readers++;
        notifyAll();
    }

    public synchronized void releaseReadLock() throws InterruptedException {
        while (isWriteLocked || readers == 0) {
            wait();
        }
        readers--;
        notifyAll();
    }

    public synchronized void acquireWriteLock() throws InterruptedException {
        while (isWriteLocked || readers > 0) {
            wait();
        }
        isWriteLocked = true;
        notifyAll();
    }

    public synchronized void releaseWriteLock() throws InterruptedException {
        while (!isWriteLocked || readers > 0) {
            wait();
        }
        isWriteLocked = false;
        notifyAll();
    }
}

The differences are I check if it is legal to release locks before return from release methods.
I assume that the given answer assumes that a thread will release the lock only after a lock is acquired. Otherwise, there is a bug that when a thread release the lock, the readers-- makes int readers to be negative, and that does not make sense.

Course: Java Multithreading for Senior Engineering Interviews - Learn Interactively
Lesson: ReadWrite Lock - Java Multithreading for Senior Engineering Interviews

#2

You can use do that without wait methods also

    public synchronized void releaseReadLock() throws InterruptedException {
        readers=Math.max(--readers,0);
        notifyAll();
    }