I am trying to use Buffer package and copy the following code from Buffer documentation.
package main
import (
"bytes"
"fmt"
"os"
)
func main() {
var b bytes.Buffer // A Buffer needs no initialization.
b.Write([]byte("Hello "))
fmt.Fprintf(&b, "world!")
b.WriteTo(os.Stdout)
}
Why do Buffer here, not to be initialize?
As you can see here Buffer consists just of some ints, the buf slice and some arrays. All of them need no initialization, since go has zero values.
You can read more about slices and arrays and how they work here.
fmt.Fprintf(&b, "world!"). When I watch Fprintf documentation, it does not expect a pointer.
It is initialized. When you do not specifically initialize a variable, go will initialize it to its zero value. That means all the internal fields of a bytes.Buffer gets the value 0, or similar for the relevant types (e.g. nil for pointers).
The authors then implemented bytes.Buffer so all values being 0 is a meaningful starting point(It means an empty buffer), so programmers doesn't need to explicitly initialize it in order to start using a Buffer.
This due to the fact that when you call the Fprintf method, the bytes.Buffer.Write method is implicitely called, and as per the doc:
Write appends the contents of p to the buffer, growing the buffer as needed.
If you look at the source code, Write calls the grow function: func (b *Buffer) grow(n int) int.
This function recognizes that the buffer is empty, because it assumes that an empty buffer has 0 values for its internal fields, which is actually how a bytes.Buffer structure is initialized by default, just like every structure in go.
