在 Scala 中处理多个期货答案

【问题标题】：Dealing with Multiple Futures in Scala在 Scala 中处理多个期货
【发布时间】：2014-08-27 03:39:59
【问题描述】：

我目前有一个基于Future[String] 更新 Redis 表的方法。

  def update(key: String, timeStamp: Long, jsonStringF: Future[String], redisClient: RedisClient) = {

    jsonStringF.map { jsonString =>
      val historyKey = "history." + key
      val tempKey = "temp." + key

      val tran = redisClient.transaction()
      tran.zadd(historyKey, (timeStamp, jsonString))
      tran.del(tempKey)
      val f = tran.exec()
      f.onComplete {
        case Success(suc) => dlogger.info(s"Updated $historyKey and deleted $tempKey successfully ....")
        case Failure(ex) => dlogger.warn(s"Error updating tables", ex)
      }
    }
  }

现在我有两个Future[String]（jsonStringF1 和jsonStringF2），我想更新两个不同的表。

  def update(key: String, timeStamp: Long, jsonStringF1: Future[String], jsonStringF2: Future[String], redisClient: RedisClient) = {
 ....

}

我想在jsonStringF2 中用String 更新另一个表("another." + key)。我怎样才能做到这一点？

更新：下面的代码是否正确？

  def update(tableKey: String, timeStamp: Long,  jsonStringF1: Future[String], jsonStringF2: Future[String], redisClient: RedisClient) =
  {

    for {
      a <- jsonStringF1
      t <- jsonStringF2

      historyKey = "history." + tableKey
      anotherKey = "another." + tableKey + 

      tempKey = "temp." + tableKey
      tran = redisClient.transaction()
      _ = tran.zadd(historyKey, (timeStamp, a))
      _ = tran.zadd(anotherKey, (timeStamp, t))
      _ = tran.del(tempKey)
      f = tran.exec()
    } yield ()
  }

【问题讨论】：

标签： scala future

【解决方案1】：

您可以使用 for 循环，如您所述

def update(tableKey: String, timeStamp: Long,  jsonStringF1: Future[String], jsonStringF2:Future[String], redisClient: RedisClient) = {
    for {
        a <- jsonStringF1
        t <- jsonStringF2
    } yield {
        val historyKey = "history." + tableKey
        val anotherKey = "another." + tableKey

        val tran = redisClient.transaction()
        tran.zadd(historyKey, (timeStamp, a))
        tran.zadd(anotherKey, (timeStamp, t))
        tran.del(tempKey)
        tran.exec()
    }
}

作为 for 的替代方案，您还可以使用 scala/async (https://github.com/scala/async) 并像这样编写代码

def update(tableKey: String, timeStamp: Long,  jsonStringF1: Future[String], jsonStringF2:Future[String], redisClient: RedisClient) = {
    async {
        val a = await(jsonStringF1)
        val t = await(jsonStringF2)

        val historyKey = "history." + tableKey
        val anotherKey = "another." + tableKey

        val tran = redisClient.transaction()
        tran.zadd(historyKey, (timeStamp, a))
        tran.zadd(anotherKey, (timeStamp, t))
        tran.del(tempKey)
        tran.exec()
    }
}

这也将是非阻塞的。异步有一点优势，因为

异步块被编译为单个匿名类，而不是每个生成器所需的每个闭包都有单独的匿名类。

【讨论】：

【解决方案2】：

您可以使用for 收集多个期货：

val f1: Future[String] = ...
val f2: Future[String] = ...

for {
  a <- f1
  b <- f2
} {
  // Code to execute when both f1 and f2 complete
  // a and b are available here
}

【讨论】：

【解决方案3】：

由于您只有 2 个期货，您可以压缩它们：

def update(tableKey: String, timeStamp: Long,  jsonStringF1: Future[String], jsonStringF2:Future[String], redisClient: RedisClient) = {

    val jsons = jsonStringF1.zip(jsonStringF2)
    jsons map {
      case (a, t) => 
        val historyKey = "history." + tableKey
        val anotherKey = "another." + tableKey

        val tran = redisClient.transaction()
        tran.zadd(historyKey, (timeStamp, a))
        tran.zadd(anotherKey, (timeStamp, t))
        tran.del(tempKey)
        tran.exec()
    }
}

【讨论】：