Variable names
Go manages the visibility with lowercase for unexported
and uppercase for exported
. These concepts are valid in the context of a package, so if you declare a variable in a package X
with the following code:
package X
var foo int // unexported
var Bar int // exported
You will be able to call the variable foo
and Bar
while you're inside that package, but if you are in another package, you will have access only to the Bar variable.
Package names
Regarding the package names is not recommended to use the underscore _
in the naming, and usually a single word, although it's just a recommendation, so follow them if it makes sense for you.
Also, the package names are used to reference the variables, constants, and types of a package, so having:
package player
type Player struct {}
and calling it from our main package will look like this:
package main
import "player"
myPlayer := player.Player{}
So we see that there is too much player verbosity, so maybe it's better to rethink the name of the package or the name of the struct.
Don't use Getters
If we have a type like:
type Player struct {
Name string
Dorsal int
}
don't create getters for each field:
func (p Player) GetName() string {
return p.Name
}
func (p Player) GetDorsal() int {
return p.Dorsal
}
Access directly to each field like Player.Name
or Player.Dorsal
or if you need a function do it without the Get
prefix.
Interfaces
The convention for the interfaces says that they should be a single method and named with the method name + -er
suffix or similar.
For example, if we want to create an interface for our defenders or strikers in our Team, following the convention we would do it like this:
type Defender interface {
Defend() error
}
type Striker interface {
Strike() error
}
Interface composition
When you have a single-methods interface like the ones above, you can create another interface, composing it by those interfaces. So our Player interface will be:
type Player interface {
Defender
Striker
}
CamelCase vs snake_case
In Go Camel Case it's the one most used. It's strange to find a snake case in any variable or constant name. In packages I already mentioned that is not recommended but sometimes you need to do it.
Constants
Constants are like variables, but with a constant value. They should not be declared all in upper case like you do in other languages. The unique upper case you should add is the first letter if you want to export it (like variables) or not.
const Minutes = 90
const minutes = 90
const MINUTES = 90 // not recommended
Grouping variables and constants
If you are declaring more than one variable or more than one constant, you can do it as we did here, or you can group them to improve the readability of your code:
const (
parts = 2
players = 22
)
var (
fizz = "fizz"
buzz = "buzz"
)
This is super useful when you're writing a new function and at the beginning you declare the variables grouped like:
func fn() error {
var (
x int
y byte
z string
)
// Logic here
return nil
}
Enums
In Go the concept of enums
doesn't exist but they can be replicated using constants and a type iota
which is an integer enumerator. To do it, you can define your enum as:
type Position int
const (
Goalkeeper Position = iota
Defender
Mid
Striker
)
With that, you can work with the constants as usual, and they will be of type Position, instead of iota (int).
Declaring it like this is enough most of the time but it introduces a possible bug in your code. The zero-value of the iota is 0, the same as the int, and we know that Go has zero-values by default.
So if you define a new position but you don't initialize it with any of the constants, Since the default position is Goalkeeper, your new default position will be Goalkeeper, and perhaps it is not the behaviour you want the application to have.
You can solve that in 2 ways. The first one is by simply adding +1 to the initial value, so the 0 value will not be any of the ones you defined.
type Position int
const (
Goalkeeper Position = iota + 1
Defender
Mid
Striker
)
That's OK, but it still has the problem of, which Position is the value 0. To solve this, the common solution is to define the zero value as an Unknown or Default or similar value like:
type Position int
const (
Unknown Position = iota
Goalkeeper
Defender
Mid
Striker
)
With these 2 tricks, you are now ready to work with your Position enum. What if you want to print those? They are int, so you will see just numbers... maybe that's not something you care about, but if you do, here is a trick that you can do to work with strings.
Go has the Stringer() interface with its method String() string
that you can satisfy with any type you want. Taking advantage of that we can do this:
func (p Position) String() string {
switch p{
case Goalkeeper:
return "Goalkeeper"
case Defender:
return "Defender"
case Mid:
return "Mid"
case Striker:
return "Striker"
}
return "Unknown"
}
Constructors
Sometimes the zero value of a type isn't enough and you want to initialize your type. For that, you need to use constructors.
package player
type Player struct {
Name string
Dorsal int
}
func NewPlayer(name string, dorsal int) *Player {
return &Player{
Name: name,
Dorsal: dorsal,
}
}
With the NewPlayer
function we initialize the Player values with the information provided as an argument instead of the zero values that would be "" and 0.
Following the naming recommendation that we described before, we can refactor this into something more idiomatic, because now to call that constructor from main would look like this:
package main
import "player"
myPlayer := player.NewPlayer("gopher", 666)
Better if we do a small refactor like:
package player
type Player struct {
Name string
Dorsal int
}
func New(name string, dorsal int) *Player {
return &Player{
Name: name,
Dorsal: dorsal,
}
}
so in our main the readability will be improved as:
package main
import "player"
myPlayer := player.New("gopher", 666)
Multiple return values
In Go, any function or method can return multiple values, from 0 to N. The error
usually is returned as the latest value.
// 2 values return
func (Player) Shoot() (int, error) {
return 99, nil
}
// single value return
func (Player) Refill() error {
// do logic
return
}
// no return value
func (p *Player) Substitute(name string) {
p.Name = name
}
Don't use panic
Most of the time simply returning errors is enough, but what if the error is unrecoverable and the program cannot continue? For that case exists the panic
.
The recommendation is to not use panics, but if you need to, the convention says to call the function with the prefix Must
. Let's see an example:
// without panic
func ParseFiles() error {
// do logic
if err != nil {
// handle error
return err
}
}
//with panic
func MustParseFiles() {
// do logic
if err != nil {
// handle error
panic()
}
}
Order declaring functions for a specific type
When you have a new type, as we saw before, most of the time you're going to create a function that acts as a constructor, and probably a list of methods for that type, maybe exported, maybe unexported.
What's the best order to declare each function? Let's see what the convention says.
// first, declare the type
type Shape strut {
x int
y int
color int
}
// next, the constructor
func NewShape(x, y, c int) *Shape {
return &Shape{
x: x,
y: y,
color: c,
}
}
// next, the exported methods
func (s *Shape) Paint(c int) {
s.color = c
}
// last, the unexported methods
func (s *Shape) reset() {
s.x = 0
s.y = 0
s.color = 0
}
I hope everything that I have tried to explain in this post has been clear, and please if there is any part that has not been completely clear or there are parts that I have not covered that you would like me to do, leave me a comment right here or through my social networks that you have on my profile and I will be happy to respond.
Top comments (5)
Most of them are the conventions that must be followed, not 'Tips and tricks'
You're right, but not everyone does it, that's why I thought I'd put it in another post where it's clear and better understood. Maybe I didn't get it and I apologize if it bothered you. I hope to do better in the next post.
Thanks for your feedback.
I agree, I would recommend everyone using
github.com/mgechev/revive
staticcheck.io/
Either directly or via golangci-lint
I agree that golangci-lint can help check a lot of these points for people, but I think it's also important for people to know why the linter is complaining about something.
Thanks for sharing the links as well.
Of course, I recommend it as a source of empowering yourself with Go. I learned a lot by looking at the issues reported on my repositories