Structs in Go

Each unnamed struct type literal starts with a struct keyword which is followed by a sequence of field definitions enclosed in a {}. Generally, each field definition is composed of a name and a type. The number of fields of a struct type can be zero.

The following is an unnamed struct type literal:

struct {
	title  string
	author string
	pages  int
}

The above struct type has three fields. The types of the two fields title and author are both string. The type of the pages field is int.

Some articles also call fields as member variables.

Consecutive fields with the same type can be declared together.

struct {
	title, author string
	pages         int
}

The size of a struct type is the sum of the sizes of all its field types plus the number of some padding bytes. The padding bytes are used to align the memory addresses of some fields. We can learn padding and memory address alignments in a later article.

The size of a zero-field struct type is zero.

A tag may be bound to a struct field when the field is declared. Field tags are optional, the default value of each field tag is a blank string. The syntax allows either string literal forms for field tags. However, in practice, struct filed tags should present as key-value pairs, and each tag should present as raw string literals (`...`), whereas each value in a tag should present as interpreted string literals ("..."). For example:

struct {
	Title  string `json:"title" myfmt:"s1"`
	Author string `json:"author,omitempty" myfmt:"s2"`
	Pages  int    `json:"pages,omitempty" myfmt:"n1"`
	X, Y   bool   `myfmt:"b1"`
}

Note, the tags of the X and Y fields in the above example are identical (though using field tags as this way is a bad practice).

We can use the reflection way to inspect field tag information.

The purpose of each field tag is application dependent. In the above example, the field tags can help the functions in the encoding/json standard package to determine the field names in JSON texts, in the process of encoding struct values into JSON texts or decoding JSON texts into struct values. The functions in the encoding/json standard package will only encode and decode the exported struct fields, which is why the first letters of the field names in the above example are all upper cased.

It is not a good idea to use field tags as comments.

Unlike C language, Go structs don't support unions.

All above shown struct types are unnamed. In practice, named struct types are more popular.

Only exported fields of struct types shown up in a package can be used in other packages by importing the package. We can view non-exported struct fields as private/protected member variables.

The field tags and the order of the field declarations in a struct type matter for the identity of the struct type. Two unnamed struct types are identical only if they have the same sequence of field declarations. Two field declarations are identical only if their respective names, their respective types and their respective tags are all identical. Please note, two non-exported struct field names from different packages are always viewed as two different names.

A struct type can't have a field of the struct type itself, neither directly nor recursively.

Given a struct type S whose underlying type is struct{x int; y bool}, the zero value of S can be represented by the following two variants of struct composite literal forms:

1. S{0, false}. In this variant, no field names are present but all field values must be present by the field declaration orders.
2. S{x: 0, y: false}, S{y: false, x: 0}, S{x: 0}, S{y: false} and S{}. In this variant, each field item is optional and the order of the field items is not important. The values of the absent fields will be set as the zero values of their respective types. But if a field item is present, it must be presented with the FieldName: FieldValue form. The order of the field items in this form doesn't matter. The form S{} is the most used zero value representation of type S.

If S is a struct type imported from another package, it is recommended to use the second form, to maintain compatibility. Consider the case where the maintainer of the package adds a new field for type S, this will make the use of first form invalid.

Surely, we can also use the struct composite literals to represent non-zero struct value.

For a value v of type S, we can use v.x and v.y, which are called selectors (or selector expressions), to represent the field values of v. v is called the receiver of the selectors. Later, we call the dot . in a selector as the property selection operator.

An example:

package main

import (
	"fmt"
)

type Book struct {
	title, author string
	pages         int
}

func main() {
	book := Book{"Go 101", "Tapir", 256}
	fmt.Println(book) // {Go 101 Tapir 256}

	// Create a book value with another form.
	// All of the three fields are specified.
	book = Book{author: "Tapir", pages: 256, title: "Go 101"}

	// None of the fields are specified. The title and
	// author fields are both "", pages field is 0.
	book = Book{}

	// Only specify the author field. The title field
	// is "" and the pages field is 0.
	book = Book{author: "Tapir"}

	// Initialize a struct value by using selectors.
	var book2 Book // <=> book2 := Book{}
	book2.author = "Tapir Liu"
	book2.pages = 300
	fmt.Println(book2.pages) // 300
}

The last , in a composite literal is optional if the last item in the literal and the closing } are at the same line. Otherwise, the last , is required. For more details, please read line break rules in Go.

var _ = Book {
	author: "Tapir",
	pages: 256,
	title: "Go 101", // here, the "," must be present
}

// The last "," in the following line is optional.
var _ = Book{author: "Tapir", pages: 256, title: "Go 101",}

When a struct value is assigned to another struct value, the effect is the same as assigning each field one by one.

func f() {
	book1 := Book{pages: 300}
	book2 := Book{"Go 101", "Tapir", 256}

	book2 = book1
	// The above line is equivalent to the
	// following lines.
	book2.title = book1.title
	book2.author = book1.author
	book2.pages = book1.pages
}

Two struct values can be assigned to each other only if their types are identical or the types of the two struct values have an identical underlying type (considering field tags) and at least one of the two types is an unnamed type.

The fields of an addressable struct are also addressable. The fields of an unaddressable struct are also unaddressable. The fields of unaddressable structs can't be modified. All composite literals, including struct composite literals are unaddressable.

Example:

package main

import "fmt"

func main() {
	type Book struct {
		Pages int
	}
	var book = Book{} // book is addressable
	p := &book.Pages
	*p = 123
	fmt.Println(book) // {123}

	// The following two lines fail to compile, for
	// Book{} is unaddressable, so is Book{}.Pages.
	/*
	Book{}.Pages = 123
	p = &(Book{}.Pages) // <=> p = &Book{}.Pages
	*/
}

Note that the precedence of the property selection operator . in a selector is higher than the address-taking operator &.

Generally, only addressable values can take addresses. But there is a syntactic sugar in Go, which allows us to take addresses on composite literals. A syntactic sugar is an exception in syntax to make programming convenient.

For example,

package main

func main() {
	type Book struct {
		Pages int
	}
	// Book{100} is unaddressable but can
	// be taken address.
	p := &Book{100} // <=> tmp := Book{100}; p := &tmp
	p.Pages = 200
}

In the following example, for simplicity, (*bookN).pages could be written as bookN.pages. In other words, bookN is dereferenced in the simplified selectors.

package main

func main() {
	type Book struct {
		pages int
	}
	book1 := &Book{100} // book1 is a struct pointer
	book2 := new(Book)  // book2 is another struct pointer
	// Use struct pointers as structs.
	book2.pages = book1.pages
	// This last line is equivalent to the above line.
	// In other words, if the receiver is a pointer,
	// it will be implicitly dereferenced.
	(*book2).pages = (*book1).pages
}

Values of two struct types S1 and S2 can be converted to each other's types, if S1 and S2 share the identical underlying type (by ignoring field tags). In particular if either S1 or S2 is an unnamed type and their underlying types are identical (considering field tags), then the conversions between the values of them can be implicit.

Given struct types S0, S1, S2, S3 and S4 in the following code snippet,

• values of type S0 can't be converted to the other four types, and vice versa, because the corresponding field names are different.
• two values of two different types among S1, S2, S3 and S4 can be converted to each other's type.

In particular,

• values of the type denoted by S2 can be implicitly converted to type S3, and vice versa.
• values of the type denoted by S2 can be implicitly converted to type S4, and vice versa.

But,

• values of the type denoted by S2 must be explicitly converted to type S1, and vice versa.
• values of type S3 must be explicitly converted to type S4, and vice versa.

package main

type S0 struct {
	y int "foo"
	x bool
}

// S1 is an alias of an unnamed type.
type S1 = struct {
	x int "foo"
	y bool
}

// S2 is also an alias of an unnamed type.
type S2 = struct {
	x int "bar"
	y bool
}

// If field tags are ignored, the underlying
// types of S3(S4) and S1 are same. If field
// tags are considered, the underlying types
// of S3(S4) and S1 are different.
type S3 S2 // S3 is a defined (so named) type
type S4 S3 // S4 is a defined (so named) type

var v0, v1, v2, v3, v4 = S0{}, S1{}, S2{}, S3{}, S4{}
func f() {
	v1 = S1(v2); v2 = S2(v1)
	v1 = S1(v3); v3 = S3(v1)
	v1 = S1(v4); v4 = S4(v1)
	v2 = v3; v3 = v2 // the conversions can be implicit
	v2 = v4; v4 = v2 // the conversions can be implicit
	v3 = S3(v4); v4 = S4(v3)
}

In fact, two struct values can be assigned (or compared) to each other only if one of them can be implicitly converted to the type of the other.

Anonymous struct types are allowed to be used as the types of the fields of another struct type. Anonymous struct type literals are also allowed to be used in composite literals.

An example:

var aBook = struct {
	// The type of the author field is
	// an anonymous struct type.
	author struct {
		firstName, lastName string
		gender              bool
	}
	title string
	pages int
}{
	author: struct { // an anonymous struct type
		firstName, lastName string
		gender              bool
	}{
		firstName: "Mark",
		lastName: "Twain",
	},
	title: "The Million Pound Note",
	pages: 96,
}

Generally, for better readability, it is not recommended to use anonymous struct type literals in composite literals.

• About Go 101 - why this book is written.
• Acknowledgments

• An Introduction of Go - why Go is worth learning.
• The Go Toolchain - how to compile and run Go programs.

• Become Familiar With Go Code
  • Introduction of Source Code Elements
  • Keywords and Identifiers
  • Basic Types and Their Value Literals
  • Constants and Variables - also introduces untyped values and type deductions.
  • Common Operators - also introduces more type deduction rules.
  • Function Declarations and Calls
  • Code Packages and Package Imports
  • Expressions, Statements and Simple Statements
  • Basic Control Flows
  • Goroutines, Deferred Function Calls and Panic/Recover

• Go Type System
  • Go Type System Overview - a must read to master Go programming.
  • Pointers
  • Structs
  • Value Parts - to gain a deeper understanding into Go values.
  • Arrays, Slices and Maps - first-class citizen container types.
  • Strings
  • Functions - function types and values, including variadic functions.
  • Channels - the Go way to do concurrency synchronizations.
  • Methods
  • Interfaces - value boxes used to do reflection and polymorphism.
  • Type Embedding - type extension in the Go way.
  • Type-Unsafe Pointers
  • Generics - use and read composite types
  • Reflections - the reflect standard package.

• Some Special Topics
  • Line Break Rules
  • More About Deferred Function Calls
  • Some Panic/Recover Use Cases
  • Explain Panic/Recover Mechanism in Detail - also explains exiting phases of function calls.
  • Code Blocks and Identifier Scopes
  • Expression Evaluation Orders
  • Value Copy Costs in Go
  • Bounds Check Elimination

• Concurrent Programming
  • Concurrency Synchronization Overview
  • Channel Use Cases
  • How to Gracefully Close Channels
  • Other Concurrency Synchronization Techniques - the sync standard package.
  • Atomic Operations - the sync/atomic standard package.
  • Memory Order Guarantees in Go
  • Common Concurrent Programming Mistakes

• Memory Related
  • Memory Blocks
  • Memory Layouts
  • Memory Leaking Scenarios

• Some Summaries
  • Some Simple Summaries
  • nil in Go
  • Value Conversion, Assignment and Comparison Rules
  • Syntax/Semantics Exceptions
  • Go FAQ 101

• More Go Related Topics