删除切片中的每个元素

Question

I'm getting some unexpected behavior when I try to loop through a slice and remove every element in sequence to print the remaining elements, using the suggested Delete method from SliceTricks. For example, when I try and loop through a slice that contains the letters [A B C], I expect the output to be [B C], [A C], [A B] in that order:

Method 1

package main

import "fmt"

func main() {
    a := []string {"A", "B", "C"}
    for i, _ := range a {
        fmt.Println(append(a[:i], a[i+1:]...))
    }
}

However, the output here is surprising to me. It outputs [B C] three times.

I did eventually get my expected behavior by doing the following:

Method 2

package main

import "fmt"

func main() {
    a := []string {"A", "B", "C"}
    for i, _ := range a {
        result := make([]string, 0)
        result = append(result, a[:i]...)
        result = append(result, a[i+1:]...)
        fmt.Println(result)
    }
}

What is causing the unexpected behavior in method 1 and is there a better way to achieve this other than method 2?

Answer 1

The "what" is because append(slice, elems...) is actually updating the slice with the new elements. It returns a "new" slice only because it may have had to relocate the original slice due to memory reallocation. So, in your example code, you are actually changing the contents of a with each call to append. (A good overview is in the "Append: An Example" section of this golang blog post).

As to "how", my attempt is:

package main

import "fmt"

func main() {
    a := []string {"A", "B", "C"}
    for i, _ := range a {
        result := make([]string, i, len(a)-1)
        copy(result, a[:i])
        result = append(result, a[i+1:]...)
        fmt.Println(result)
    }
}

DEMO

Initializing result with a length and capacity, as well as using copy instead of two appends attempts to reduce the number of memory reallocations required.