How to implement resizable arrays in Go

Question

I come from a C++ background and I'm used to using the std::vector class for things like this. Lets assume i want a dynamic array of these:

type a struct {
    b int
    c string
}

What is the standard way of doing this?

A snippet would be very useful

Answer 1

you might also be able to make do with a slice. which is an array that knows its current length. And can have a separate current length and maximum capacity. Note the values passed for initial size and capacity do not have to be constants so you can create a function which builds and returns slices of different lengths based on its parameters.

The up side is that a slice []Int can just be indexed like an array, and will return ints when used in this way.

The downside is that it will not automatically grow byound its stated capacity. Effective Go has an example of how you would go about handling reallocation.

the code would be

type mytype struct {
   a, b int
}




func main() {

  sl := make([]mytype, 10, 50) //slice of 10 items, max capacity 50 these do not have to be constant expressions.
  sl[0] = mytype{1,2}
   //...
  for i, value := range sl {
  // ... do stuff with value
  }
}

Answer 2

A Go Slice contains three elements: data, length, and capacity.

s := make([]int, 0, 10)

The variable s is a slice of ints with a length of 0 and a capacity of 10. The built-in len() and cap() functions allow you to get the length and capacity of a slice:

len(s) == 0
cap(s) == 10

To increase the length of a slice, simply re-slice:

s = s[0:5]
// len(s) == 5
// cap(s) == 10

To decrease the length, you can take a sub-slice:

s = s[0:1]
// len(s) == 1

There are some shorter ways to invoke make():

a := make([]int, 10) 
// len(a) == cap(a) == 10

b := make([]int)
// len(b) == cap(b) == 0

That's all well and good, but what if you need to increase the length of a slice beyond its capacity? To do that, you need to allocate a new slice and copy the contents of the old slice to the new one. (The function "copy" is another built-in.)

t := make([]int, len(s), 20)
copy(t, s)

The Effective Go document takes this example a bit further, implementing an Append function that appends one slice to another, resizing it if necessary.

Slices are backed by arrays; when you make() a slice of a specific capacity, an array of that capacity is allocated in the background. The slice effectively becomes a "smart pointer" to that array. If you pass that slice (or a subslice of that slice) to another function, it is passed as a pointer to that same array. This makes sub-slices very cheap to create - it's the allocation of the backing array that is expensive.

The Go standard library includes a number of container packages - vector, for instance - that eliminate the need to manually manage slices. Use slices for speed, and more elaborate container classes for convenience. (Saying that, I still use slices for most things.)

You may be wondering why you need to go to all this trouble. After all, a lot of languages provide dynamically resized arrays as primitives. The reason for this is tied to Go's philosophy. The language designers don't presume to know what the appropriate allocation policy is for your program; instead they give you the tools you need to build your own data structures.