Go Notes

This is a page of notes on a particular technology. It's meant to describe the subject in a succient manner and comes from a collection of important points derived from a concentrated study. It should be useful as reference or getting a high-level understanding of the subject quickly.

This is an introduction to the Go programming language.


  • Each Go file declares a package.
  • Go depends on a specific file structure.
  • Go style is enforced by the go fmt tool. Go expects certain styling to compile.
  • All Go programs have a func main() in a package ‘main’.
  • Uppercase methods are public and exportable, lowercase methods are private.
  • Go constructs strings from runes, which are unicode characters. Unicode characters are also useable in the language itself (ex: function names).

Documentation is at golang.org. It’s also available using the go doc tool.

go doc <module> <function>


Go programs depend on a specific file structure.

  • src/ contains source files which declare packages.
  • src/<package>/<file>.go is a file within a package.
  • bin/ contains compiled binaries from Go packages.


Go source is compiled and executed from a GOPATH.

cd ./<project>
export GOPATH=`pwd`
go install <project>


Go has static types with hinting. You can explicitly declare types, or you can have Go figure it out using :=.

The following are equivalent:

var message string
message = "Hello"

// or

var message string = "Hello"

// or

message := "Hello"

:= will also allow you to redeclare some variables, as long as part of the declaration statement includes new variables.

old_var := 3
new_var, old_var := 3, 4

Outside functions, const blocks declare global constants and var blocks declare global variables. Capitalized constants and variables export them for use by outside modules.

const (
  foo = "bar"

var (
  fizz = "buzz"

An _ can be used to throw away unneeded information.

var foo, _ = myFunc()

Go will error on unused variables and variables that attempt to be declared more than once.

All variables, when declared, are initialized to their “zero value,” which depends on the type (for example, booleans get ‘False’ and integers get 0).


Types can be declared using type functions.

var number = uint(9)  // 9 is an argument to the uint type function.

Type functions can also do some conversions. Each conversion creates a copy.

my_string := "This is a test"
my_byte_slice := []byte(my_string)
back_to_string = string(my_byte_slice)
  • iota can create enumerations and is an auto incrementing value.

    var answer1 = iota
    var answer2
    fmt.Printf("%d %d")   // Prints '1 2'.
    var answer1 = iota*4
    var answer2
    fmt.Printf("%d %d")   // Prints '4 8'.
  • float32 and float64 are floating point types.

    var pi float64 = 3.1415926
    pi = float64(3.14159)
  • uint8 and unint64 are integers.
  • bool is a boolean. ! is negation.
  • Also byte.
  • Also error – the built-in Go error type.
  • The empty value in Go is nil.

Defined Types

Structs can be used to define types.

type something struct {
  foo string
  bar int

s := &something(foo: "test")
s.bar = 3

Both examples below will instantiate a struct with zero-values.

s := &something{}
s := new(something)

Types and the type keyword can also be used to alias and extend existing types with methods.

type SuperIntSlice []int
s := &SuperIntSlice{1, 2, 3, 4, 5, 6}


Strings can be indexed using array brackets and :.

astr := "This is my string.\n"


Looping over a string rune by rune:

for i, r := range astr {
  fmt.Printf("%d %c\n", i, r)

The range keyword allows k/v iteration.

for _, r := range astr {

Get string length with len().

Strings in back quotes, `, are interpreted literally. Double quotes, "", will interpret special characters, like \n.

Arrays and Slices

Go handles normal array behavior with arrays and slices.

Declaring arrays can be done bounded or unbounded.

unbounded := [...]string{"foo", "bar"}
bounded := [2]string{"fizz", "buzz"}

Slices are a specific sections of arrays. They are defined using empty [].

my_slice := []string{"the", "quick", "brown", "fox"}

Slices can be formed on the fly from arrays or slices. The first value is the starting index (inclusive), the second value is the ending index (exclusive).

foo = unbounded_slice[1:2]  // Includes only unbounded_slice[1]
bar = unbounded_slice[:3]   // Includes the first three items.

Arrays are fixed in memory and are only passed between functions by copying, whereas slices are passed by reference. Slices are more performant when passed between functions because it doesn’t copy data, and allows array data to be directly modified outside of its declaration scope.

Dynamic slices can be created using the make function, which allocates memory.

words := make([]string, 4)      // Initializes with four empty strings.
words := make([]string, 0, 4)   // Intiializes with no entries, but a max
                                // capacity of four.

words = append(words, "Added string")
  • len() will get length and cap() will show capacity.
  • append() will automatically expand slice capacity if it’s exceeded.
  • Slices can be explicitly copied using copy().

Attempting to access sections of a slice or array that do not exist will cause “out of bounds” runtime errors.

Slices have no equality property, so they cannot be compared, except to nil.


Maps can be instantiated using the map keyword, either by make or by explicit definition.

my_map := make(map[key_type]value_type)

my_other_map := map[key_type]value_type{
    "foo": "bar",

Accessing an element of the map that doesn’t exist will return the zero type. An error can be explicitly thrown by checking for a return value.

foo, ok := my_map["bar"]    // "bar" doesn't exist.
if !ok {
  // This was an errorneous lookup.
} else {
  // The lookup worked.

Iterating over a map uses range. Maps are not sorted by default, so results will come out in arbitrary order.

for month, days := range days_in_months {
  fmt.Printf("There are %d days in %s.\n", days, month)

Adding a value to a map just requires an assignment.

my_map["foo"] = "bar"

Items can be deleted from maps. Deletion will silently pass if the key does not exist.

delete(my_map, "foo")
delete(my_map, "foo")  // Does not delete, does not error.

Keys can be any type that has equality.


Go can use if statements for error handling. Variables declared within if statements are only accessible within the conditional block.

if readCount, err = module.Format("Hello.\n"); err != nil {
  // Respond to an error.
} else if readCount == 0 {
  fmt.Printf("Read nothing!")
}else {

Go also has switch statements.

switch {
case foo == "bar":
case foo == "bizz":

Switch statements do not have a necessary break and do not fall-through. Instead, the fallthrough keyword will cause cases to collapse.

case foo == "bar":

You can also switch on a particular variable.

r = someRandomCharacter()

switch r {
case "a", "e", "i", "o", "u":
  fmt.Printf("r turned out to be a vowel.")
  fmt.Printf("r is weird!")


The for loop covers all cases.

An infinite loop:

for {
  fmt.Printf("Hello, world.\n")

A constrained loop

counter := 1
for counter < 100 {
  counter += 1

// or ...

for counter := 0; counter < 10; counter++ {
  fmt.Printf("Foo bar.\n")

Go does have post-fix updating, foo++, but it isn’t a statement and is only valid inside loop clauses.

Loops can simultaneously use assignments and multiple conditions.


Calling a function takes this form:

<package>.<function>([args ...])

A function definition looks like this:

func name(param1 type, param2, param3 shared_type) (return_type1, return_type2) {
  ... statements ...

  return foo, bar
  • Functions can return multiple things.
  • Parameters declare their type after the variable in the argument list.
  • Return type parentheses () are optional if there is only one return type.

The defer keyword queues up actions to be taken after a function exits. Statements which are defered execute in stack order. Useful for function clean-up.

func printer(msg []byte) error {
  f, err := os.Create("helloworld.txt")
  defer f.Close()

  if err == nil {

  return err

Return values in Go can also be defined at execution time for shorthand returns.

func tester(msg []byte) (e error) {
  _, e = fmt.Printf("%s\n", msg)
  return  // Automatically returns `e`.

Go supports variable number of parameters with the ... parameter.

func tester(fmt string, msgs ...string) {
  for _, msg := range msgs {
    fmt.Printf(fmt, msg)

In general, pass-by-reference is accomplished by using pointers. Otherwise, Go usually uses pass-by-copy.


Go has errors which return from functions and (very rarely used) panics to handle extreme events.

func my_printer(msg string) error {
  _, err := fmt.Printf("%s\n", msg)
  return err

if err := my_printer("Hello, world."); err != nil {
  // The function returned an error.

By default, errors are using the built-in standard library.

Defining errors can be done using error formatting, Errorf or the errors library:

  • Errorf can manipulate the standard error to say something unique.

      fmt.Errorf("This is an error message!")
  • errors can be used to define custom error types. Downstream behavior can check the type of error and respond accordingly.

      var (
        errorTerribleThing = errors.New("Something terrible happened!")
      func do_something() error {
        return erroTerribleThing

panic and recover in Go can stop execution or return stack traces if something happens that is not recoverable.