### 傳統異常處理的局限性
~~~
try {
1024 / 0
} catch {
case e: Throwable => e.printStackTrace()
}
~~~
用try-catch的模式,異常必須在拋出的時候馬上處理.
然而在分布式計算中,我們很可能希望將異常集中到一起處理,來避免需要到每臺機器上單獨看錯誤日志的窘態.
~~~
val seq = Seq(0, 1, 2, 3, 4)
//seq: Seq[Int] = List(0, 1, 2, 3, 4)
val seqTry = seq.map(x => Try {
20 / x
})
//seqTry: Seq[scala.util.Try[Int]] = List(Failure(java.lang.ArithmeticException: devide by zero),Success(20), Success(10), Success(6), Success(5))
val succSeq = seqTry.flatMap(_.toOption)
//succSeq: Seq[Int] = List(20, 10, 6, 5) Try可以轉換成Option
val succSeq2 = seqTry.collect {
case Success(x) => x
}
//succSeq2: Seq[Int] = List(20, 10, 6, 5) 和上一個是一樣的
val failSeq: Seq[Throwable] = seqTry.collect {
case Failure(e) => e
}
//failSeq: Seq[Throwable] = List(java.lang.ArithmeticException: devide by zero)
~~~
Try實例可以序列化,并且在機器間傳送.
