Try-catch是用来帮助异常处理的。这意味着它将在某种程度上帮助我们的系统变得更加健壮:尝试从意外事件中恢复。

我们怀疑在执行和指令(发送消息)时可能会发生一些事情,因此它被包含在try中。如果发生了几乎意想不到的事情,我们可以做些什么:我们编写捕获。我不认为我们调用只是为了记录异常。我认为catch块是为了给我们从错误中恢复的机会。

现在,假设我们从错误中恢复,因为我们可以修复错误。如果能再试一次,那就太好了:

try{ some_instruction(); }
catch (NearlyUnexpectedException e){
   fix_the_problem();
   retry;
}

这将很快陷入永恒循环,但让我们假设fix_the_problem返回true,然后重试。假设在Java中没有这样的东西,你将如何解决这个问题?你认为解决这个问题的最佳设计代码是什么?

这就像一个哲学问题,因为我已经知道Java不直接支持我所要求的内容。


当前回答

你可以从jcabi-aspect(我是一名开发人员)使用AOP和Java注释:

@RetryOnFailure(attempts = 3, delay = 5)
public String load(URL url) {
  return url.openConnection().getContent();
}

您还可以使用@Loggable和@LogException注释。

其他回答

强制性的“企业级”解决方案:

public abstract class Operation {
    abstract public void doIt();
    public void handleException(Exception cause) {
        //default impl: do nothing, log the exception, etc.
    }
}

public class OperationHelper {
    public static void doWithRetry(int maxAttempts, Operation operation) {
        for (int count = 0; count < maxAttempts; count++) {
            try {
                operation.doIt();
                count = maxAttempts; //don't retry
            } catch (Exception e) {
                operation.handleException(e);
            }
        }
    }
}

并呼吁:

OperationHelper.doWithRetry(5, new Operation() {
    @Override public void doIt() {
        //do some stuff
    }
    @Override public void handleException(Exception cause) {
        //recover from the Exception
    }
});

这些答案基本上都是一样的。我的也是,但我喜欢这种形式

boolean completed = false;
Throwable lastException = null;
for (int tryCount=0; tryCount < config.MAX_SOME_OPERATION_RETRIES; tryCount++)
{
    try {
        completed = some_operation();
        break;
    }
    catch (UnlikelyException e) {
        lastException = e;
        fix_the_problem();
    }
}
if (!completed) {
    reportError(lastException);
}

https://onlinegdb.com/a-7RsL1Gh

    public void doSomething() throws Exception{
      final int MAX_TRIES = 10;
      int count = 0;
      
      while(count++ < MAX_TRIES){
         try{ 
            System.out.println("trying");
            causeIssue(count); // throws error/exception till count 2
            System.out.println("trying successful");
            break; // break on success
         } catch (Exception e){
           System.out.println("caught, logging Exception:" + count);
         } catch (Error e){
           System.out.println("caught, logging Error:" + count);
         }
      }
    }

输出:

trying
caught, logging Error:1
trying
caught, logging Error:2
trying
trying successful

这里是我的解决方案类似于其他人可以包装一个函数,但允许您获得函数返回值,如果它成功。

    /**
     * Wraps a function with retry logic allowing exceptions to be caught and retires made.
     *
     * @param function the function to retry
     * @param maxRetries maximum number of retires before failing
     * @param delay time to wait between each retry
     * @param allowedExceptionTypes exception types where if caught a retry will be performed
     * @param <V> return type of the function
     * @return the value returned by the function if successful
     * @throws Exception Either an unexpected exception from the function or a {@link RuntimeException} if maxRetries is exceeded
     */
    @SafeVarargs
    public static <V> V runWithRetriesAndDelay(Callable<V> function, int maxRetries, Duration delay, Class<? extends Exception>... allowedExceptionTypes) throws Exception {
        final Set<Class<? extends Exception>> exceptions = new HashSet<>(Arrays.asList(allowedExceptionTypes));
        for(int i = 1; i <= maxRetries; i++) {
            try {
                return function.call();
            } catch (Exception e) {
                if(exceptions.contains(e.getClass())){
                    // An exception of an expected type
                    System.out.println("Attempt [" + i + "/" + maxRetries + "] Caught exception [" + e.getClass() + "]");
                    // Pause for the delay time
                    Thread.sleep(delay.toMillis());
                }else {
                    // An unexpected exception type
                    throw e;
                }
            }
        }
        throw new RuntimeException(maxRetries + " retries exceeded");
    }

解决这个问题的一个简单方法是在while循环中包装try/catch并维护一个计数。通过这种方式,您可以根据其他变量检查计数,同时维护故障日志,从而防止出现无限循环。这不是最精致的解决方案,但它是可行的。