Channels can be closed. If a channel is closed, no values can be sent on it. A channel is open the moment you create it, e.g. c := make(chan int, 5).

You may want to close a channel, to indicate completion. That is to say, when the goroutines are completed it may be the channel has no reason to be open.

Example

Closing channel

The line c := make(chan int, 5) creates a buffered channel of capacity 5. A buffered channel can store multiple values without it being received.

Then data is sent to the channels with the lines c <- 5 and c <- 3.

The channel is closed with the function close(channel).

package main

func main() {
    c := make(chan int, 5)
    c <- 5
    c <- 3
    close(c)
}
    

If you sent data after the channel is closed, like c <- 1 after the close(c) call, it will throw this error:

    $ go run example.go
    panic: send on closed channel
    
    goroutine 1 [running]:
    main.main()
        /home/linux/golang/example.go:11 +0x9b
        exit status 2

Receive data

If a channel is closed, you can still read data. But you cannot send new data into it. This program reads both before and after closing the channel, which works. It’s closed only for sneding, meaning a line like c <- 9 won’t work after close(c).

package main

import "fmt"

func main() {
    c := make(chan int, 5)

    c <- 5
    c <- 3

    fmt.Println(<-c)
    close(c)
    fmt.Println(<-c)
}