981
工業級應用中關於異常封裝的一些感悟
Java的異常體係想必大家都熟記於心的,那麼在日常的編碼過程中,大家又是如何實施的呢?針對checked和unchecked exception大家的理解又有多少呢?
幾乎所有的書本上都給出了一個conclusion,如果日常事務能從異常中恢複,那麼建議用checked exception,如果不能recover from,那麼就是unchecked的了.
大家想過沒有,這裏的恢複指的是什麼?真的不能恢複嗎?堆棧,Thread又是如何表現的呢?
This exception That bug,靜觀皆自得.好了,開場白過後,這裏,主要總結一下自己對Java Exception的理解,也歡迎大家留言,發表自己的心得~
(1) 一種簡單的checked,unchecked exception的理解是,將checked exception作為業務異常去使用(關注它,則處理它,不關注,re-throw),可以使用errorCode之類的進行wrap,不要迷戀recover from,自己去實地感知一下,就全了然了;
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
result.cancel(true);//Future<Boolean> result = pool.submit(new Callable<Boolean>() {...});
} catch (ExecutionException e) {
if (e.getCause() instanceof RuntimeException)
throw new RuntimeException(e);
else if (e.getCause() instanceof Error)
throw new Error(e.getCause());
else
throw new IllegalStateException("Not unchecked!", e.getCause());
}
(4) 建立分級的異常監控,預警機製.(使用腳本語言,諸如perl,python等主動,被動兼施)；
(5) 有技巧的"吞掉"異常.不按套路出牌的人，永遠讓人難以防備，關於這一點，可以研習一下jetty的continuation異步模型；
(6) 嚐試使用Scala（或者Java7+的多重捕獲）,與Java不同，scala裏全然是unchecked exception。另外通過模式匹配，異常處理邏輯，代碼可讀性也更好.如：
def tryBreakable[T](op: =>T) = new TryBlock[T] {
def catchBreak(onBreak: =>T) = try {
op
} catch {
case ex: BreakControl =>
if (ex ne breakException) throw ex
onBreak
}
}
補充篇：
最近設計OpenAPI的時候，由於要和前端進行交互，後端用的Webx3框架，部分
異常采用了其Pipeline處理，部分采用了編碼映射處理。具體講，就是錯誤碼與狀態碼進行編排，前端按需所示（框架層麵摟住5XX，4XX狀態碼，應用層麵對其它碼進行個性化展示。當然也可以采用位運算取而代之碼編排）。示例代碼如下：
/**
* @author vongosling 2013-2-1
* @see com.openapi.exception.ErrorCode
*/
public enum CommonErrorMapper {
E_10000("10000", "500", "Unknown error"),
E_10001("10001", "500", "System error"),
E_10002("10002", "503", "Service unavailable "),
E_10003("10003", "500", "Remote service error"),
E_10004("10004", "400", "Illegal request"),
E_10005("10005", "500", "Param error, see doc for more info"),
E_10006("10006", "400", "Request records over limit "),
E_10007("10007", "400", "IP limit "),
E_10008("10008", "400", "Insufficient app permissions"),
E_10009("10009", "404", "Request service not found "),
E_10010("10010", "400", "IP requests out of rate limit"),
E_10011("10011", "400", "User requests out of rate limit"),
E_10012("10012", "400", "Access token does not exist "),
E_10013("10013", "400", "Access token expired "),
E_10015("10015", "400", "Request body length over limit ");
private String httpStatus = "";
private String errorCode = "";
private String errorMessage = "";
public String getHttpStatus() {
return httpStatus;
}
public String getErrorCode() {
return errorCode;
}
public String getErrorMessage() {
return errorMessage;
}
private CommonErrorMapper(String errorCode, String httpStatus, String errorMessage) {
this.httpStatus = httpStatus;
this.errorCode = errorCode;
this.errorMessage = errorMessage;
}
然後在全局攔截點進行處理，實例代碼如下：
if (null != result.getErrors()) {
CommonErrorMapper mapper = CommonErrorMapper.valueOf("E_"+ result.getErrors().get(0).getCode());//注意異常兼容，互斥等原則
if (null != mapper) {
response.setErrorCode(CommonErrorMapper.getErrorCode());
response.setHttpStatus(CommonErrorMapper.getHttpStatus());
response.setContent(CommonErrorMapper.getErrorMessage());
} else {
response.setErrorCode(CommonErrorMapper.E_10001.getErrorCode());
response.setHttpStatus(CommonErrorMapper.E_10001.getHttpStatus());
response.setContent(result.getErrors());
}
} else {
response.setHttpStatus("200");
response.setContent(result.getContent());
response.setMetadata( responseObject.getMetaData());
}
return response; }前端返回RPC結果為:
jcb1360050602232(
{
content: "Resubmit error.",
errorCode: "10026",
metadata: null,
httpStatus: "500"
})
剩下的就一目了然了吧～
下麵給出一些比較有啟發意義的異常處理文獻:
1.https://tutorials.jenkov.com/java-exception-handling/index.html
2.https://onjava.com/pub/a/onjava/2003/11/19/exceptions.html
3. https://code.google.com/p/guava-libraries/wiki/ThrowablesExplained
最後更新：2017-04-02 06:52:05