您如何理解commons-lang3中“ LazyInitializer”的代码

Question

这几天，我正在研究apache-commons-lang3的源代码。我无法理解对“ LazyInitializer”类的方法“ get（）”的评论。

那么“ LazyInitializer.get（）”的两个实现之间有什么区别？

apache-commons-lang3中的默认实现：

public abstract class LazyInitializer<T> implements ConcurrentInitializer<T> {

    private static final Object NO_INIT = new Object();

    @SuppressWarnings("unchecked")
    /** Stores the managed object. */
    private volatile T object = (T) NO_INIT;

    /**
     * Returns the object wrapped by this instance. On first access the object
     * is created. After that it is cached and can be accessed pretty fast.
     *
     * @return the object initialized by this {@code LazyInitializer}
     * @throws ConcurrentException if an error occurred during initialization of
     * the object
     */
    @Override
    public T get() throws ConcurrentException {
        // use a temporary variable to reduce the number of reads of the
        // volatile field
        T result = object;

        if (result == NO_INIT) {
            synchronized (this) {
                result = object;
                if (result == NO_INIT) {
                    object = result = initialize();
                }
            }
        }

        return result;
    }

    /**
     * Creates and initializes the object managed by this {@code
     * LazyInitializer}. This method is called by {@link #get()} when the object
     * is accessed for the first time. An implementation can focus on the
     * creation of the object. No synchronization is needed, as this is already
     * handled by {@code get()}.
     *
     * @return the managed data object
     * @throws ConcurrentException if an error occurs during object creation
     */
    protected abstract T initialize() throws ConcurrentException;
}

我的实现：

... omited

    @Override
    public T get() throws ConcurrentException {
        if (result != null) {
            return result;
        }

        synchronized (this) {
            if (result != null){
                return result;
            }
            result = initialize();
        }

        return result;
    }

我不明白为什么要这样设计或有什么好处？