在Scala中同步var [英] Synchronize on a var in Scala

查看:95
本文介绍了在Scala中同步var的处理方法,对大家解决问题具有一定的参考价值,需要的朋友们下面随着小编来一起学习吧!

问题描述

问题是在几个页面的文本之间搜索感叹号并发,一旦任何线程找到它,所有其他线程应该停止搜索。

The problem is to search among a few pages of text for an exclamation point concurrently and as soon as any of the threads finds it, all the other threads should stop searching.

代码:

  object AntiVolatile {
    val pages = for (i <- 1 to 15) yield new Page("!Na" * rand.nextInt(1000) + " Batman!", -1)
    var found = Some(false)

    def run(): Unit = {
      for (p <- pages) yield thread {
        var i = 0
        var foundInThread = found.get
        while (i < p.txt.length && !foundInThread)
          if (p.txt(i) == '!') {
            found.synchronized {
              found match {
                case Some(true) => foundInThread = true
                case Some(false) => {
                  p.position = i
                  found = Some(true)
                  Thread.sleep(1)
                }
                case _ =>
              }
            }
          } else i += 1
        // if still not found, wait for another thread to find it.
        def wait(): Unit = {
          found match {
            case Some(false) => wait()
            case _ =>
          }
        }
        wait()
        log(s"results: ${pages.map(_.position)}")
      }
    }
  }

似乎工作正常:

Thread-29: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-27: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-28: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-26: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-30: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-31: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-32: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-25: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-33: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-34: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-39: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-38: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-37: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-36: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)
Thread-35: results: Vector(0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1)

但我意识到找到一个常量实例,因为它会重新分配一个新的选项对象。 (为什么代码实际工作?)

But then I realized that found is not a constant instance because it's reassign to a new Option object later. (Why does the code actually work?)

所以我想出了一个修复:

So I came up with a fix:

  object AntiVolatile {
    case class Found(var isFound: Boolean)
    val pages = for (i <- 1 to 15) yield new Page("!Na" * rand.nextInt(1000) + " Batman!", -1)
    val found = Found(false)

    def run(): Unit = {
      for (p <- pages) yield thread {
        var i = 0
        var foundInThread = found.isFound
        while (i < p.txt.length && !foundInThread)
          if (p.txt(i) == '!') {
            found.synchronized {
              found match {
                case Found(true) => foundInThread = true
                case Found(false) => {
                  p.position = i
                  found.isFound = true
                  Thread.sleep(1)
                }
                case _ =>
              }
            }
          } else i += 1
        // if still not found, wait for another thread to find it.
        def wait(): Unit = {
          found match {
            case Found(false) => wait()
            case _ =>
          }
        }
        wait()
        log(s"results: ${pages.map(_.position)}")
      }
    }
  }

这两个版本看起来都是一样的,为什么?我希望在第一个版本中出现一些错误。

These two versions seem to behave the same, why? I would expect some bug cropping up in the first version.

链接到github repo: https://github.com/kindlychung/learnConcurrentScala/blob/master/src/main/scala/org/learningconcurrency/ch2/Ch2.scala p>

Link to github repo: https://github.com/kindlychung/learnConcurrentScala/blob/master/src/main/scala/org/learningconcurrency/ch2/Ch2.scala

推荐答案

如果你对学习并发感兴趣,或者如果你正在解决一个实际的问题,还不完全清楚。话说回来,我想假设你正试图解决这个问题。

It is not entirely clear if you are interested in learning concurrency or if you are solving an actual problem. With that being said, I am going to assume that you are attempting to solve the problem.

为什么不使用Futures?

Why not use Futures?

import java.util.concurrent.TimeUnit

import scala.concurrent.duration.Duration
import scala.concurrent.{ExecutionContext, Await, Future}
import scala.util.Random
import ExecutionContext.Implicits.global

object Main extends App {

  case class Page(number: Int, text: String)

  val pages = for (i <- 1 to 15) yield Page(i, "!Na" * Random.nextInt(1000) + " Batman! ")

  val searchFutures = pages.map { p => Future {
    val position = p.text.indexOf("!")
    s"Exclamation mark found on page ${p.number} at position: $position"
  }}

  val firstCompleted = Future.firstCompletedOf(searchFutures)

  val result = Await.result(firstCompleted, Duration(5, TimeUnit.SECONDS))
  println(result)

}

这篇关于在Scala中同步var的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!

查看全文
登录 关闭
扫码关注1秒登录
发送“验证码”获取 | 15天全站免登陆