关于并行流中的线程安全问题

现象

在我们使用java8中的parallel流时可能会遇到线程安全问题，如下面的例子：

    @Test
    public void test1() {
        System.out.println("==== test ArrayList ====");
        for (int i = 0; i < 10; i++) {
            // the initial capacity is must
            List<Integer> list = new ArrayList<>(10000);
            IntStream.range(0, 10000)
                    .boxed()
                    .parallel()
                    .forEach(data -> list.add(data));
            System.out.println("NO." + i + ": " + list.size());
        }

        // test HashMap
        System.out.println("==== test HashMap ====");
        for (int i = 0; i < 10; i++) {
            Map<Integer, Integer> map = new HashMap<>();
            IntStream.range(0, 10000)
                    .boxed()
                    .parallel()
                    .forEach(data -> map.put(data, data));
            System.out.println("NO." + i + ": " + map.size());
        }
    }

运行结果：

==== test ArrayList ====
NO.0: 8634
NO.1: 6443
NO.2: 6229
NO.3: 3624
NO.4: 2894
NO.5: 3888
NO.6: 3405
NO.7: 3722
NO.8: 3532
NO.9: 3850
==== test HashMap ====
NO.0: 8355
NO.1: 7732
NO.2: 8986
NO.3: 8796
NO.4: 8620
NO.5: 8873
NO.6: 7203
NO.7: 7882
NO.8: 8382
NO.9: 7346

解决方案

1. 使用java8提供的收集器

@Test
    public void test2() {
        System.out.println("==== test List ====");
        for (int i = 0; i < 10; i++) {
            List<Integer> list = IntStream.range(0, 10000)
                    .boxed()
                    .parallel()
                    .collect(Collectors.toList());
            System.out.println("NO." + i + ": " + list.size());
        }

        // test HashMap
        System.out.println("==== test Map ====");
        for (int i = 0; i < 10; i++) {
            Map<Integer, Integer> map = IntStream.range(0, 10000)
                    .boxed()
                    .parallel()
                    .collect(Collectors.toMap(
                            data -> data,
                            data -> data
                            )
                    );
            System.out.println("NO." + i + ": " + map.size());
        }
    }

结果

==== test List ====
NO.0: 10000
NO.1: 10000
NO.2: 10000
NO.3: 10000
NO.4: 10000
NO.5: 10000
NO.6: 10000
NO.7: 10000
NO.8: 10000
NO.9: 10000
==== test Map ====
NO.0: 100000
NO.1: 100000
NO.2: 100000
NO.3: 100000
NO.4: 100000
NO.5: 100000
NO.6: 100000
NO.7: 100000
NO.8: 100000
NO.9: 100000

2. 使用Collections.synchronizedList()和Collections.synchronizedMap()分别包装我们的ArrayList和HashMap

    @Test
    public void test3() {
        System.out.println("==== test ArrayList ====");
        for (int i = 0; i < 10; i++) {
            List<Integer> list = Collections
                    .synchronizedList(new ArrayList<>(10000));
            IntStream.range(0, 10000)
                    .boxed()
                    .parallel()
                    .forEach(data -> list.add(data));
            System.out.println("NO." + i + ": " + list.size());
        }

        // test HashMap
        System.out.println("==== test HashMap ====");
        for (int i = 0; i < 10; i++) {
            Map<Integer, Integer> map = Collections.synchronizedMap(new HashMap<>());
            IntStream.range(0, 10000)
                    .boxed()
                    .parallel()
                    .forEach(data -> map.put(data, data));
            System.out.println("NO." + i + ": " + map.size());
        }
    }

结果

==== test ArrayList ====
NO.0: 10000
NO.1: 10000
NO.2: 10000
NO.3: 10000
NO.4: 10000
NO.5: 10000
NO.6: 10000
NO.7: 10000
NO.8: 10000
NO.9: 10000
==== test HashMap ====
NO.0: 10000
NO.1: 10000
NO.2: 10000
NO.3: 10000
NO.4: 10000
NO.5: 10000
NO.6: 10000
NO.7: 10000
NO.8: 10000
NO.9: 10000

原因解释