Go.初探 Goroutine 与 channel

前言

Go 语言的 CSP 并发模型的实现包含两个主要组成部分:一个是 Goroutine,另一个是 channel。本文将会介绍它们的基本用法和注意事项。

Goroutine

Goroutine 是 Go 应用的基本执行单元,它是一种轻量的用户级线程,其底层是通过 coroutine(协程)去实现的并发。众所周知,协程是一种运行在用户态的用户线程,因此 Goroutine 也是被调度于 Go 程序运行时。

基本用法

语法:go + 函数/方法

通过 go 关键字 + 函数/方法 可以创建一个 Goroutine。

代码示例:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
import (
"fmt"
"time"
)

func printGo() {
fmt.Println("具名函数")
}

type G struct {
}

func (g G) g() {
fmt.Println("方法")
}

func main() {
// 基于具名函数创建 goroutine
go printGo()
// 基于方法创建 goroutine
g := G{}
go g.g()
// 基于匿名函数创建 goroutine
go func() {
fmt.Println("匿名函数")
}()
// 基于闭包创建 goroutine
i := 0
go func() {
i++
fmt.Println("闭包")
}()
time.Sleep(time.Second) // 避免 main goroutine 结束后,其创建的 goroutine 来不及运行,因此在此休眠 1 秒
}

执行结果:

1
2
3
4
5
闭包
具名函数
方法
匿名函数

当多个 Goroutine 存在时,它们的执行顺序是不固定的。因此每次打印的结果都不相同。

由代码可知,通过 go 关键字,我们可以基于 具名函数 / 方法 创建 goroutine,也可以基于 匿名函数 / 闭包 创建 goroutine。

那么 Goroutine 是如何退出的呢?正常情况下,只要 Goroutine 函数执行结束,或者执行返回,意味着 Goroutine 的退出。如果 Goroutine 的函数或方法有返回值,在 Goroutine 退出时会将其忽略

channel

channel 在 Go 并发模型中扮演者重要的角色。它可以用于实现 Goroutine 间的通信,也可以用来实现 Goroutine 间的同步。

channel 的基本操作

channel 是一种复合数据类型,声明时需要指定 channel 里元素的类型。

声明语法:var 变量名 chan 类型
通过上述代码声明一个元素类型为 string 的 channel,其只能存放 string 类型的元素。channel 是引用类型,必须初始化才能写入数据,通过 make 的方式初始化。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
import (
"fmt"
)

func main() {
var ch chan string
ch = make(chan string, 1)
// 打印 chan 的地址
fmt.Println(ch)
// 向 ch 发送 "Go" 数据
ch <- "Go"
// 从 ch 中接收数据
s := <-ch
fmt.Println(s) // Go
}

通过 ch <- xxx 可以向 channel 变量 ch 发送数据,通过 x := <- ch 可以从 channel 变量 ch 中接收数据。

带缓冲 channel 与无缓冲 channel

如果初始化 channel 时,不指定容量时,则创建的是一个无缓冲的 channel:

1
2
ch := make(chan string)

一文初探 Goroutine 与 channel
作者:陈明勇
2023-04-21
广东
本文字数:2665 字

阅读完需:约 9 分钟

一文初探 Goroutine 与 channel
作者:陈明勇

专注分享后端知识,如果本文对你有帮助,不妨点个赞,如果你是 Go 语言初学者,不妨点个关注,一起成长一起进步,如果本文有错误的地方,欢迎指出!

前言
Go 语言的 CSP 并发模型的实现包含两个主要组成部分:一个是 Goroutine,另一个是 channel。本文将会介绍它们的基本用法和注意事项。

Goroutine
Goroutine 是 Go 应用的基本执行单元,它是一种轻量的用户级线程,其底层是通过 coroutine(协程)去实现的并发。众所周知,协程是一种运行在用户态的用户线程,因此 Goroutine 也是被调度于 Go 程序运行时。

基本用法

语法:go + 函数/方法

通过 go 关键字 + 函数/方法 可以创建一个 Goroutine。

代码示例:

import (
“fmt”
“time”
)

func printGo() {
fmt.Println(“具名函数”)
}

type G struct {
}

func (g G) g() {
fmt.Println(“方法”)
}

func main() {
// 基于具名函数创建 goroutine
go printGo()
// 基于方法创建 goroutine
g := G{}
go g.g()
// 基于匿名函数创建 goroutine
go func() {
fmt.Println(“匿名函数”)
}()
// 基于闭包创建 goroutine
i := 0
go func() {
i++
fmt.Println(“闭包”)
}()
time.Sleep(time.Second) // 避免 main goroutine 结束后,其创建的 goroutine 来不及运行,因此在此休眠 1 秒
}

执行结果:

闭包
具名函数
方法
匿名函数

当多个 Goroutine 存在时,它们的执行顺序是不固定的。因此每次打印的结果都不相同。

由代码可知,通过 go 关键字,我们可以基于 具名函数 / 方法 创建 goroutine,也可以基于 匿名函数 / 闭包 创建 goroutine。

那么 Goroutine 是如何退出的呢?正常情况下,只要 Goroutine 函数执行结束,或者执行返回,意味着 Goroutine 的退出。如果 Goroutine 的函数或方法有返回值,在 Goroutine 退出时会将其忽略。

channel
channel 在 Go 并发模型中扮演者重要的角色。它可以用于实现 Goroutine 间的通信,也可以用来实现 Goroutine 间的同步。

channel 的基本操作
channel 是一种复合数据类型,声明时需要指定 channel 里元素的类型。

声明语法:var 变量名 chan 类型

通过上述代码声明一个元素类型为 string 的 channel,其只能存放 string 类型的元素。channel 是引用类型,必须初始化才能写入数据,通过 make 的方式初始化。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
import (
"fmt"
)

func main() {
var ch chan string
ch = make(chan string, 1)
// 打印 chan 的地址
fmt.Println(ch)
// 向 ch 发送 "Go" 数据
ch <- "Go"
// 从 ch 中接收数据
s := <-ch
fmt.Println(s) // Go
}

通过 ch <- xxx 可以向 channel 变量 ch 发送数据,通过 x := <- ch 可以从 channel 变量 ch 中接收数据。

带缓冲 channel 与无缓冲 channel
如果初始化 channel 时,不指定容量时,则创建的是一个无缓冲的 channel:

1
ch := make(chan string)

无缓冲的 channel 的发送与接收操作是同步的,在执行发送操作之后,对应 Goroutine 将会阻塞,直到有另一个 Goroutine 去执行接收操作,反之亦然。如果将发送操作和执行操作放在同一个 Goroutine 下进行,会发生什么操作呢?看看下述代码:

1
2
3
4
5
6
7
8
9
10
11
12
13
import (
"fmt"
)

func main() {
ch := make(chan int)
// 发送数据
ch <- 1 // fatal error: all goroutines are asleep - deadlock!
// 接收数据
n := <-ch
fmt.Println(n)
}

程序运行之后,会在 ch <- 处得到 fatal error,提示所有的 Goroutine 处于休眠状态,也就是死锁了。为避免这种情况,我们需要将 channel 的发送操作和接收操作放到不同的 Goroutine 中执行。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
import (
"fmt"
)

func main() {
ch := make(chan int)
go func() {
// 发送数据
ch <- 1
}()
// 接收数据
n := <-ch
fmt.Println(n) // 1
}

由上述例子可以得出结论:无缓冲 channel 的发送与接收操作,一定要放在两个不同的 Goroutine 中进行,否则会发生 deadlock 形象。

如果指定容量,则创建的是一个带缓冲的 channel:

1
2
3
4
5
由上述例子可以得出结论:无缓冲 channel 的发送与接收操作,一定要放在两个不同的 Goroutine 中进行,否则会发生 deadlock 形象。



如果指定容量,则创建的是一个带缓冲的 channel:

有缓冲的 channel 与无缓冲的 chennel 有所区别,执行发送操作时,只要 channel 的缓冲区未满,Goroutine 不会挂起,直到缓冲区满时,再向 channel 执行发送操作,才会导致 Goroutine 挂起。代码示例:

1
2
3
4
5
6
7
8
func main() {
ch := make(chan int, 1)
// 发送数据
ch <- 1

ch <- 2 // fatal error: all goroutines are asleep - deadlock!
}

声明 channel 的只发送类型和只接收类型

既能发送又能接收的 channel

  • 既能发送又能接收的 channel
    1
    2
    ch := make(chan int, 1)

    通过上述代码获得 channel 变量,我们可以对它执行发送与接收的操作。
  • 只接收的 channel
    1
    2
    ch := make(<-chan int, 1)

    通过上述代码获得 channel 变量,我们只能对它进行接收操作。
  • 只发送的 channel
    1
    ch := make(chan<- int, 1)
    通过上述代码获得 channel 变量,我们只能对它进行发送操作。

通常只发送 channel 类型和只接收 channel 类型,会被用作函数的参数类型或返回值:

1
2
3
4
5
6
7
8
func send(ch chan<- int) {
ch <- 1
}

func recv(ch <-chan int) {
<-ch
}

channel 的关闭

通过内置函 close(c chan<- Type),可以对 channel 进行关闭。

  • 在发送端关闭 channel

  • 在 channel 关闭之后,将不能对 channel 执行发送操作,否则会发生 panic,提示 channel 已关闭。

    1
    2
    3
    4
    5
    6
    7
    func main() {
    ch := make(chan int, 5)
    ch <- 1
    close(ch)
    ch <- 2 // panic: send on closed channel
    }

  • 管道 channel 之后,依旧可以对 channel 执行接收操作,如果存在缓冲区的情况下,将会读取缓冲区的数据,如果缓冲区为空,则获取到的值为 channel 对应类型的零值

    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    import "fmt"

    func main() {
    ch := make(chan int, 5)
    ch <- 1
    close(ch)
    fmt.Println(<-ch) // 1
    n, ok := <-ch
    fmt.Println(n) // 0
    fmt.Println(ok) // false
    }

  • 如果通过 for-range 遍历 channel 时,中途关闭 channel 则会导致 for-range 循环结束。