Golang对类型数组中的切片字段的引用

Question

I'm a Python programmer enjoying Go. The thing that trips me up must often is the use of references; I have mastered it (mostly) but then on occasion something baffles me, such as this one.

I have a simple type ('Fixture'):

type Fixture struct {
    Probabilities      *[]float64
}

If I populate the Probabilities field for a single instance of this type, everything is fine:

c := appengine.NewContext(r)
f := Fixture{}
p := []float64{}
p = append(p, 0.5)
p = append(p, 0.2)
p = append(p, 0.3)
f.Probabilities = &p
c.Infof("%v", *f.Probabilities)

2013/03/19 07:37:36 INFO: [0.5 0.2 0.3]

However if I try and populate this field for an array of these types, the code compiles but the field value is always nil:

c := appengine.NewContext(r)
fixtures := []Fixture{}
f := Fixture{}
fixtures = append(fixtures, f)
for _, f := range fixtures {
    p := []float64{}
    p = append(p, 0.5)
    p = append(p, 0.2)
    p = append(p, 0.3)
    f.Probabilities = &p
}
for _, f := range fixtures {
    // c.Infof("%v", *f.Probabilities) // causes error
    c.Infof("%v", f.Probabilities)
}

2013/03/19 07:37:41 INFO: <nil>

I guess I'm not understanding how arrays/slices work, particularly with respect to references. Can anyone point out where I'm going wrong ?

Thanks!

Answer 1

In the range statement

for _, f := range fixtures { ... }

f is a newly declared local variable of type Fixture. It's not a reference to anything. So after setting up its values, it has to be put into to the fixtures slice.

package main

import "fmt"

type Fixture struct {
        Probabilities *[]float64
}

func main() {
        fixtures := []Fixture{}
        f := Fixture{}
        fixtures = append(fixtures, f)
        for i, f := range fixtures {
                p := []float64{}
                p = append(p, 0.5)
                p = append(p, 0.2)
                p = append(p, 0.3)
                f.Probabilities = &p
                fixtures[i] = f
        }
        for _, f := range fixtures {
                fmt.Printf("%v", f.Probabilities)
        }
}

(Also here)

---

Output

&[0.5 0.2 0.3]