Java8 — asynchronous programming

Asynchronous programming

Asynchrony is simply implementing a method that lets the operation continue without waiting for a return value from the called function

Create a task and execute it

No arguments to create

1 CompletableFuture<String> noArgsFuture = new CompletableFuture<>();

Copy the code

The corresponding task is passed, with no return value

The runAsync method performs asynchronous computations in the background, but returns no value. Holds a Runnable object.

1CompletableFuture noReturn = CompletableFuture.runAsync((a)->{

2    //Execute logic with no return value

3});

Copy the code

The corresponding task is passed in with a return value

And now we see that it’s returned CompletableFuture

where T is the type of value that you want to return. The Supplier

is a simple functional interface.

1CompletableFuture<String> hasReturn = CompletableFuture.supplyAsync(new Supplier<String> () {

2    @Override

3    public String get() {

4        return "hasReturn";

5    }

6});

Copy the code

Lambda expressions can be used at this point to make the above logic clearer

1CompletableFuture<String> hasReturnLambda = CompletableFuture.supplyAsync(TestFuture::get);

2

3private static String get() {

4    return "hasReturnLambda";

5}

Copy the code

Get the return value

An asynchronous task also has a return value, and when we want to use the return value of an asynchronous task, we can call the CompletableFuture get() block until an asynchronous task has returned a value.

Let’s modify the get() method above to pause it for ten seconds.

 1private static String get() {

 2    System.out.println("Begin Invoke getFuntureHasReturnLambda");

 3    try {

 4        Thread.sleep(10000);

 5    } catch (InterruptedException e) {

 6

 7    }

 8    System.out.println("End Invoke getFuntureHasReturnLambda");

 9    return "hasReturnLambda";

10}

Copy the code

And then call

1public static void main(String[] args) throws ExecutionException, InterruptedException {

2    CompletableFuture<String> funtureHasReturnLambda = (CompletableFuture<String>) getFuntureHasReturnLambda();

3    System.out.println("Main Method Is Invoking");

4    funtureHasReturnLambda.get(a);

5    System.out.println("Main Method End");

6}

Copy the code

As you can see in the output, the current thread is blocked only when the get() method is called.

1Main Method Is Invoking

2Begin Invoke getFuntureHasReturnLambda

3End Invoke getFuntureHasReturnLambda

4Main Method End

Copy the code

Custom return value

Instead of waiting for an asynchronous task to return a value, we can customize the return value at any time by calling the complete() method.

 1CompletableFuture<String> funtureHasReturnLambda = (CompletableFuture<String>) getFuntureHasReturnLambda();

 2System.out.println("Main Method Is Invoking");

 3new Thread(()->{

 4    System.out.println("Thread Is Invoking ");

 5    try {

 6        Thread.sleep(1000);

 7        funtureHasReturnLambda.complete("custome value");

 8    } catch (InterruptedException e) {

 9        e.printStackTrace();

10    }

11    System.out.println("Thread End ");

12}).run();

13String value = funtureHasReturnLambda.get(a);

14System.out.println("Main Method End value is "+ value);

Copy the code

We can see that the output is the value of the new thread, of course, because our asynchronous method is set to wait 10 seconds. If the asynchronous method waits 1 second and the new thread waits 10 seconds, then the output value is the value of the asynchronous method.

1Main Method Is Invoking

2Begin Invoke getFuntureHasReturnLambda

3Thread Is Invoking 

4Thread End 

5Main Method End value is custome value

Copy the code

Perform asynchronous tasks sequentially

What if you have an asynchronous task whose completion depends on the completion of a previous asynchronous task? Is the get() method called to get the return value and then executed? It’s a bit of a hassle to write this, but CompletableFuture gives us a way to do what we need to do in order to do some asynchronous tasks. ThenApply, thenAccept, and thenRun. The difference between these three methods.

The method name Whether the return value of the previous task is available Whether there is a return value
thenApply Can get There are
thenAccept Can get There is no
thenRun Do not get There is no

So generally thenAccept and thenRun are used at the end of the call chain. Let’s take a look at a real example.

 1//thenApply retrieves the return value of the previous task, as well as the return value

 2CompletableFuture<String> seqFutureOne = CompletableFuture.supplyAsync(()-> "seqFutureOne");

 3CompletableFuture<String> seqFutureTwo = seqFutureOne.thenApply(name -> name + " seqFutureTwo");

 4System.out.println(seqFutureTwo.get());

 5

 6

 7//thenAccept retrieves the return value of the previous task, but has no return value

 8CompletableFuture<Void> thenAccept = seqFutureOne

 9        .thenAccept(name -> System.out.println(name + "thenAccept"));

10System.out.println("-- -- -- -- -- -- -- -- -- -- -- -- --");

11System.out.println(thenAccept.get());

12

13//thenRun does not get the return value of the previous task, and there is no return value

14System.out.println("-- -- -- -- -- -- -- -- -- -- -- -- --");

15CompletableFuture<Void> thenRun = seqFutureOne.thenRun(() -> {

16    System.out.println("thenRun");

17});

18System.out.println(thenRun.get());

Copy the code

The following information is returned

1seqFutureOne seqFutureTwo

2seqFutureOnethenAccept

3-------------

4null

5-------------

6thenRun

7null

Copy the code

ThenApply and thenApplyAsync

We can see that all three methods have an Async suffix, such as thenApplyAsync. So what’s the difference between methods with Async and methods without the suffix? So let’s compare thenApply to thenApplyAsync, and everything else is the same.

The difference between the two methods is who performs the task. If thenApplyAsync is used, a different thread is executed from ForkJoinPool.commonPool(). If thenApply is used, If the supplyAsync method executes very fast, thenApply is the main thread, or if it executes very slowly, it is the same as the supplyAsync thread. Let’s take a look at how fast supplyAsync executes using sleep.

 1//thenApply and thenApplyAsync

 2System.out.println("-- -- -- -- -- -- -- -- -- -- -- -- --");

 3CompletableFuture<String> supplyAsyncWithSleep = CompletableFuture.supplyAsync(()->{

 4    try {

 5        Thread.sleep(10000);

 6    } catch (InterruptedException e) {

 7        e.printStackTrace();

 8    }

 9    return "supplyAsyncWithSleep Thread Id : " + Thread.currentThread();

10});

11CompletableFuture<String> thenApply = supplyAsyncWithSleep

12        .thenApply(name -> name + "------thenApply Thread Id : " + Thread.currentThread());

13CompletableFuture<String> thenApplyAsync = supplyAsyncWithSleep

14        .thenApplyAsync(name -> name + "------thenApplyAsync Thread Id : " + Thread.currentThread());

15System.out.println("Main Thread Id: "+ Thread.currentThread());

16System.out.println(thenApply.get());

17System.out.println(thenApplyAsync.get());

18System.out.println("-------------No Sleep");

19CompletableFuture<String> supplyAsyncNoSleep = CompletableFuture.supplyAsync(()->{

20    return "supplyAsyncNoSleep Thread Id : " + Thread.currentThread();

21});

22CompletableFuture<String> thenApplyNoSleep = supplyAsyncNoSleep

23        .thenApply(name -> name + "------thenApply Thread Id : " + Thread.currentThread());

24CompletableFuture<String> thenApplyAsyncNoSleep = supplyAsyncNoSleep

25        .thenApplyAsync(name -> name + "------thenApplyAsync Thread Id : " + Thread.currentThread());

26System.out.println("Main Thread Id: "+ Thread.currentThread());

27System.out.println(thenApplyNoSleep.get());

28System.out.println(thenApplyAsyncNoSleep.get());

Copy the code

We can see that the output is

1-------------

2Main Thread Id: Thread[main,5,main]

3supplyAsyncWithSleep Thread Id : Thread[ForkJoinPool.com monPool - worker - 1, 5, the main]------thenApply Thread Id : Thread[ForkJoinPool.com monPool - worker - 1, 5, the main]

4supplyAsyncWithSleep Thread Id : Thread[ForkJoinPool.com monPool - worker - 1, 5, the main]------thenApplyAsync Thread Id : Thread[ForkJoinPool.com monPool - worker - 1, 5, the main]

5-------------No Sleep

6Main Thread Id: Thread[main,5,main]

7supplyAsyncNoSleep Thread Id : Thread[ForkJoinPool.com monPool - worker - 2, 5, the main]------thenApply Thread Id : Thread[main,5,main]

8supplyAsyncNoSleep Thread Id : Thread[ForkJoinPool.com monPool - worker - 2, 5, the main]------thenApplyAsync Thread Id : Thread[ForkJoinPool.com monPool - worker - 2, 5, the main]

Copy the code

If the supplyAsync method is slow, thenApply is executing on the same thread as the supplyAsync method. If the supplyAsync method is fast, thenApply is executing on the same thread as the Main method.

Combination CompletableFuture

There are two ways to combine two CompletableFutures together

  1. thenCompose(): The second action is performed when the first task is complete
  2. thenCombine(): Operations are performed only when both asynchronous tasks are complete

The usage thenCompose ()

Let’s define two asynchronous tasks, assuming that the second scheduled task needs the return value of the first.

1public static CompletableFuture<String> getTastOne(){

2    return CompletableFuture.supplyAsync((a)-> "topOne");

3}

4

5public static CompletableFuture<String> getTastTwo(String s){

6    return CompletableFuture.supplyAsync((a)-> s + " topTwo");

7}

Copy the code

We write using the thenCompose() method

1CompletableFuture<String> thenComposeComplet = getTastOne().thenCompose(s -> getTastTwo(s));

2System.out.println(thenComposeComplet.get());

Copy the code

The output is

1topOne  topTwo

Copy the code

If you remember the thenApply() method from earlier, you might think that this method could do something similar with thenApply().

1//thenApply

2CompletableFuture<CompletableFuture<String>> thenApply = getTastOne()

3        .thenApply(s -> getTastTwo(s));

4System.out.println(thenApply.get().get());

Copy the code

But we find that the return value is a nested return type, and getting the final return value requires two get() calls

The usage thenCombine ()

For example, we need to calculate the sum of the return values of two asynchronous methods. The sum operation can only be calculated if the values obtained by two asynchronous methods, so we can use thenCombine() method for calculation.

1CompletableFuture<Integer> thenComposeOne = CompletableFuture.supplyAsync((a) -> 192);

2CompletableFuture<Integer> thenComposeTwo = CompletableFuture.supplyAsync((a) -> 196);

3CompletableFuture<Integer> thenComposeCount = thenComposeOne

4        .thenCombine(thenComposeTwo, (s, y) -> s + y);

5System.out.println(thenComposeCount.get());

Copy the code

Only when both thenComposeOne and thenComposeTwo are complete will the callback function passed to thenCombine be called.

Combine multiple CompletableFutures

Above we combine two CompletableFutures using thenCompose() and thenCombine(). What if we want to combine any number of CompletableFutures? The following two methods can be combined.

  • allOf(): Waiting for allCompletableFutureThe callback function will be run only after that
  • anyOf(): Just one of themCompletableFutureWhen done, the callback function is executed. Note that other tasks are not executed.

Let’s demonstrate the use of both methods

 1//allOf()

 2CompletableFuture<Integer> one = CompletableFuture.supplyAsync((a) -> 1);

 3CompletableFuture<Integer> two = CompletableFuture.supplyAsync((a) -> 2);

 4CompletableFuture<Integer> three = CompletableFuture.supplyAsync((a) -> 3);

 5CompletableFuture<Integer> four = CompletableFuture.supplyAsync((a) -> 4);

 6CompletableFuture<Integer> five = CompletableFuture.supplyAsync((a) -> 5);

 7CompletableFuture<Integer> six = CompletableFuture.supplyAsync((a) -> 6);

 8

 9CompletableFuture<Void> voidCompletableFuture = CompletableFuture.allOf(one, two, three, four, five, six);

10voidCompletableFuture.thenApply(v->{

11    return Stream.of(one,two,three,four, five, six)

12            .map(CompletableFuture::join)

13            .collect(Collectors.toList());

14}).thenAccept(System.out::println);

15

16CompletableFuture<Void> voidCompletableFuture1 = CompletableFuture.runAsync((a) -> {

17    try {

18        Thread.sleep(1000);

19    } catch (Exception e) {

20

21    }

22    System.out.println("1");

23});

Copy the code

We’ve defined six compleTableFutures that wait for all the completableFutures to wait for all the tasks to complete and then output their values.

The use of the anyOf ()

 1CompletableFuture<Void> voidCompletableFuture1 = CompletableFuture.runAsync((a) -> {

 2try {

 3    Thread.sleep(1000);

 4} catch (Exception e) {

 5

 6}

 7System.out.println("voidCompletableFuture1");

 8});

 9

10CompletableFuture<Void> voidCompletableFutur2 = CompletableFuture.runAsync((a) -> {

11try {

12    Thread.sleep(2000);

13} catch (Exception e) {

14

15}

16System.out.println("voidCompletableFutur2");

17});

18

19CompletableFuture<Void> voidCompletableFuture3 = CompletableFuture.runAsync((a) -> {

20try {

21    Thread.sleep(3000);

22} catch (Exception e) {

23

24}

25System.out.println("voidCompletableFuture3");

26});

27

28CompletableFuture<Object> objectCompletableFuture = CompletableFuture

29    .anyOf(voidCompletableFuture1, voidCompletableFutur2, voidCompletableFuture3);

30objectCompletableFuture.get();

Copy the code

Here we’ve defined three CompletableFutures to do some time-consuming task, at which point the first CompletableFuture will complete first. The printed result is as follows.

1voidCompletableFuture1

Copy the code

Exception handling

We learned how CompletableFuture executes asynchronously, how to combine different CompletableFutures, and how to execute completableFutures sequentially. The next important step is what to do if an exception occurs while performing an asynchronous task. So let’s write an example.

1CompletableFuture.supplyAsync((a)->{

2    //An exception occurs

3    int i = 10/0;

4    return "Success";

5}).thenRun((a)-> System.out.println("thenRun"))

6.thenAccept(v -> System.out.println("thenAccept"));

7

8CompletableFuture.runAsync((a)-> System.out.println("CompletableFuture.runAsync"));

Copy the code

As a result, we find that if an exception occurs in one of the execution chains, the next chain will not be executed, but the other CompletableFutures in the main process will still run.

1CompletableFuture.runAsync

Copy the code

exceptionally()

We can use exceptionally for exception handling

 1// Handle exceptions

 2

 3CompletableFuture<String> exceptionally = CompletableFuture.supplyAsync(() -> {

 4    // An exception occurred

 5    int i = 10 / 0;

 6    return "Success";

 7}).exceptionally(e -> {

 8    System.out.println(e);

 9    return "Exception has Handl";

10});

11System.out.println(exceptionally.get());

Copy the code

If the following output is displayed, you can find that the received value is an exception or a customized value can be returned.

1java.util.concurrent.CompletionException: java.lang.ArithmeticException: / by zero

2Exception has Handl

Copy the code

handle()

Calling the handle() method also catches exceptions and returns a custom value. Unlike the () method, the handle() method is called whether an exception occurs or not. Examples are as follows

 1System.out.println("------- is abnormal -------");

 2CompletableFuture.supplyAsync((a)->{

 3    //An exception occurs

 4    int i = 10/0;

 5    return "Success";

 6}).handle((response,e)->{

 7    System.out.println("Exception:" + e);

 8    System.out.println("Response:" + response);

 9    return response;

10});

11

12System.out.println("------- No abnormal -------");

13CompletableFuture.supplyAsync((a)->{

14    return "Sucess";

15}).handle((response,e)->{

16    System.out.println("Exception:" + e);

17    System.out.println("Response:" + response);

18    return response;

19});

Copy the code

As printed below, we can see that the handle() method is called without an exception

1------- An exception occurs -------

2Exception:java.util.concurrent.CompletionException: java.lang.ArithmeticException: / by zero

3Response:null

4------- No exception -------

5Exception:null

6Response:Sucess

Copy the code

Source code address

Refer to the article

  • Juejin. Cn/post / 684490…
  • www.jianshu.com/p/6bac52527…