Preface

The source of this book is written in mmark and is converted from the original LaTeX source.

All example code used in this book is hereby put in the public domain.

This work is licensed under the Attribution-NonCommercial-ShareAlike 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/ or send a letter to Creative Commons, 171 Second Street, Suite 300, San Francisco, California, 94105, USA.

The following people made large or small contributions to earlier versions of this book:

Adam J. Gray, Alexander Katasonov, Alexey Chernenkov, Alex Sychev, Andrea Spadaccini, Andrey Mirtchovski, Anthony Magro, Babu Sreekanth, Ben Bullock, Bob Cunningham, Brian Fallik, Cecil New, Cobold, Damian Gryski, Daniele Pala, Dan Kortschak, David Otton, Fabian Becker, Filip Zaludek, Hadi Amiri, Haiping Fan, Iaroslav Tymchenko, Jaap Akkerhuis, JC van Winkel, Jeroen Bulten, Jinpu Hu, John Shahid, Jonathan Kans, Joshua Stein, Makoto Inoue, Marco Ynema, Mayuresh Kathe, Mem, Michael Stapelberg, Nicolas Kaiser, Olexandr Shalakhin, Paulo Pinto, Peter Kleiweg, Philipp Schmidt, Robert Johnson, Russel Winder, Simoc, Sonia Keys, Stefan Schroeder, Thomas Kapplet, T.J. Yang, Uriel\(\dagger\), Vrai Stacey, Xing Xing.

"Learning Go" has been translated into (note that this used the original LaTeX source).

I hope this book is useful.

Miek Gieben, London, 2015.

This book still sees development, small incremental improvements trickle in from Github.

Miek Gieben, London, 2017.

Learning Go

Introduction

Is Go an object-oriented language? Yes and no.

The Go programming language is an open source project language to make programmers more productive.

According to the website "Go is expressive, concise, clean, and efficient". And indeed it is. My initial interest was piqued when I read early announcements about this new language that had built-in concurreny and a C-like syntax (Erlang also has built-in concurrency, but I could never get used to its syntax). Go is a compiled statically typed language that feels like a dynamically typed, interpreted language. My go to (scripting!) language Perl has taken a back seat now that Go is around.

The unique Go language is defined by these principles:

Clean and Simple
Go strives to keep things small and beautiful. You should be able to do a lot in only a few lines of code.
Concurrent
Go makes it easy to "fire off" functions to be run as very lightweight threads. These threads are called goroutines 1 in Go.
Channels
Communication with these goroutines is done, either via shared state or via channels .
Fast
Compilation is fast and execution is fast. The aim is to be as fast as C. Compilation time is measured in seconds.
Safe
Explicit casting and strict rules when converting one type to another. Go has garbage collection. No more free() in Go: the language takes care of this.
Standard format
A Go program can be formatted in (almost) any way the programmers want, but an official format exists. The rule is very simple: The output of the filter gofmt is the officially endorsed format.
Postfix types
Types are given after the variable name, thus var a int, instead of int a.
UTF-8
UTF-8 is everywhere, in strings and in the program code. Finally you can use \(\Phi = \Phi + 1\) in your source code.
Open Source
The Go license is completely open source.
Fun
Programming with Go should be fun!

As I mentioned Erlang also shares some features of Go. A notable difference between Erlang and Go is that Erlang borders on being a functional language, while Go is imperative. And Erlang runs in a virtual machine, while Go is compiled.

How to Read this Book

I've written this book for people who already know some programming languages and how to program. In order to use this book, you (of course) need Go installed on your system, but you can easily try examples online in the Go playground2. All exercises in this book work with Go 1, the first stable release of Go -- if not, it's a bug.

The best way to learn Go is to create your own programs. Each chapter therefore includes exercises (and answers to exercises) to acquaint you with the language. Each exercise is either easy, intermediate, or difficult. The answers are included after the exercises on a new page. Some exercises don't have an answer; these are marked with an asterisk.

Here's what you can expect from each chapter:

We'll look at the basic types, variables, and control structures available in the language.
Here we look at functions, the basic building blocks of Go programs.
We'll see that functions and data can be grouped together in packages. We'll also see how to document and test our packages.
We'll create our own types. We'll also look at memory allocations in Go.
We'll learn how to use interfaces. Interfaces are the central concept in Go, as Go does not support object orientation in the traditional sense.
We'll learn the go keyword, which can be used to start function in separate routines (called goroutines). Communication with those goroutines is done via channels.
Finally we'll see how to interface with the rest of the world from within a Go program. We'll see how to create files and read and write to and from them. We'll also briefly look into networking.

Official Documentation

There is a substantial amount of documentation written about Go. The Go Tutorial , the Go Tour (with lots of exercises) and the Effective Go are helpful resources. The website http://golang.org/doc/ is a very good starting point for reading up on Go3. Reading these documents is certainly not required, but it is recommended.

When searching on the internet use the term "golang" instead of plain "go".

Go comes with its own documentation in the form of a program called godoc4. If you are interested in the documentation for the built-ins, simply do this:

% godoc builtin

To get the documentation of the hash package, just:

% godoc hash

To read the documentation of fnv contained in hash, you'll need to issue godoc hash/fnv as fnv is a subdirectory of hash.

PACKAGE DOCUMENTATION

package fnv
    import "hash/fnv"

    Package fnv implements FNV-1 and FNV-1a, non-cryptographic hash
    ...

Basics

I am interested in this and hope to do something.

In this chapter we will look at the basic building blocks of the Go programming language.

Hello World

In the Go tutorial, you get started with Go in the typical manner: printing "Hello World" (Ken Thompson and Dennis Ritchie started this when they presented the C language in the 1970s). That's a great way to start, so here it is, "Hello World" in Go.

package main 1

import "fmt" 2 // Implements formatted I/O.

/* Print something */ 3
func main() { 4
	fmt.Printf("Hello, world.") 5
}

Lets look at the program line by line. This first line is just required 1. All Go files start with package <something>, and package main is required for a standalone executable.

import "fmt" says we need fmt in addition to main 2. A package other than main is commonly called a library, a familiar concept in many programming languages (see ). The line ends with a comment that begins with //.

Next we another comment, but this one is enclosed in /* */ 3. When your Go program is executed, the first function called will be main.main(), which mimics the behavior from C. Here we declare that function 4.

Finally we call a function from the package fmt to print a string to the screen. The string is enclosed with " and may contain non-ASCII characters 5.

Compiling and Running Code

To build a Go program, use the go tool. To build helloworld we just enter:

% go build helloworld.go

This results in an executable called helloworld.

% ./helloworld
Hello, world.

You can combine the above and just call go run helloworld.go.

Variables, Types and Keywords

In the next few sections we will look at the variables, basic types, keywords, and control structures of our new language.

Go is different from (most) other languages in that the type of a variable is specified after the variable name. So not: int a, but a int. When you declare a variable it is assigned the "natural" null value for the type. This means that after var a int, a has a value of 0. With var s string, s is assigned the zero string, which is "". Declaring and assigning in Go is a two step process, but they may be combined. Compare the following pieces of code which have the same effect.

var a int                           a := 15
var b bool                          b := false
a = 15
b = false

On the left we use the var keyword to declare a variable and then assign a value to it. The code on the right uses := to do this in one step (this form may only be used inside functions). In that case the variable type is deduced from the value. A value of 15 indicates an int. A value of false tells Go that the type should be bool. Multiple var declarations may also be grouped; const (see ) and import also allow this. Note the use of parentheses instead of braces:

var (
    x int
    b bool
)

Multiple variables of the same type can also be declared on a single line: var x, y int makes x and y both int variables. You can also make use of parallel assignment: a, b := 20, 16. This makes a and b both integer variables and assigns 20 to a and 16 to b.

A special name for a variable is _. Any value assigned to it is discarded (it's similar to /dev/null on Unix). In this example we only assign the integer value of 35 to b and discard the value 34: _, b := 34, 35. Declared but otherwise unused variables are a compiler error in Go.

Boolean Types

A boolean type represents the set of boolean truth values denoted by the predeclared constants true and false. The boolean type is bool.

Numerical Types

Go has most of the well-known types such as int. The int type has the appropriate length for your machine, meaning that on a 32-bit machine it is 32 bits and on a 64-bit machine it is 64 bits. Note: an int is either 32 or 64 bits, no other values are defined. Same goes for uint, the unsigned int.

If you want to be explicit about the length, you can have that too, with int32, or uint32. The full list for (signed and unsigned) integers is int8, int16, int32, int64 and byte, uint8, uint16, uint32, uint64, with byte being an alias for uint8. For floating point values there is float32 and float64 (there is no float type). A 64 bit integer or floating point value is always 64 bit, also on 32 bit architectures.

Note that these types are all distinct and assigning variables which mix these types is a compiler error, like in the following code:

package main

func main() {
	var a int
	var b int32
	b = a + a
	b = b + 5
}

We declare two different integers, a and b where a is an int and b is an int32. We want to set b to the sum of a and a. This fails and gives the error: cannot use a + a (type int) as type int32 in assignment. Adding the constant 5 to b does succeed, because constants are not typed.

Constants

Constants in Go are just that --- constant. They are created at compile time, and can only be numbers, strings, or booleans; const x = 42 makes x a constant. You can use iota 5 to enumerate values.

const (
    a = iota
    b
)

The first use of iota will yield 0, so a is equal to 0. Whenever iota is used again on a new line its value is incremented with 1, so b has a value of 1. Or, as shown here, you can even let Go repeat the use of iota. You may also explicitly type a constant: const b string = "0". Now b is a string type constant.

Strings

Another important built-in type is string. Assigning a string is as simple as:

s := "Hello World!"

Strings in Go are a sequence of UTF-8 characters enclosed in double quotes ("). If you use the single quote (') you mean one character (encoded in UTF-8) --- which is not a string in Go.

Once assigned to a variable, the string cannot be changed: strings in Go are immutable. If you are coming from C, note that the following is not legal in Go:

var s string = "hello"
s[0] = 'c'

To do this in Go you will need the following:

s := "hello"
c := []rune(s)	    1
c[0] = 'c'	        2
s2 := string(c)     3
fmt.Printf("%s\n", s2) 4

Here we convert s to an array of runes 1. We change the first element of this array 2. Then we create a new string s2 with the alteration 3. Finally, we print the string with fmt.Printf 4.

Runes

Rune is an alias for int32. It is an UTF-8 encoded code point. When is this type useful? One example is when you're iterating over characters in a string. You could loop over each byte (which is only equivalent to a character when strings are encoded in 8-bit ASCII, which they are not in Go!). But to get the actual characters you should use the rune type.

Complex Numbers

Go has native support for complex numbers. To use them you need a variable of type complex128 (64 bit real and imaginary parts) or complex64 (32 bit real and imaginary parts). Complex numbers are written as re + im\(i\), where re is the real part, im is the imaginary part and \(i\) is the literal '\(i\)' (\(\sqrt{-1}\)).

Errors

Any non-trivial program will have the need for error reporting sooner or later. Because of this Go has a builtin type specially for errors, called error. var e error creates a variable e of type error with the value nil. This error type is an interface -- we'll look more at interfaces in . For now you can just assume that error is a type just like all other types.

Operators and Built-in Functions

Go supports the normal set of numerical operators. See for lists the current ones and their relative precedence. They all associate from left to right.

Operator precedence.
Precedence Operator(s)
Highest * / % << >> & &^
`+ -
== != < <= > >=
<-
&&
Lowest ||

+ - * / and % all do what you would expect, & | ^ and &^ are bit operators for bitwise and, bitwise or, bitwise xor, and bit clear respectively. The && and || operators are logical and and logical or . Not listed in the table is the logical not : !

Although Go does not support operator overloading (or method overloading for that matter), some of the built-in operators are overloaded. For instance, + can be used for integers, floats, complex numbers and strings (adding strings is concatenating them).

Go Keywords

Let's start looking at keywords, lists all the keywords in Go.

Keywords in Go.
         
break default func interface select
case defer go map struct
chan else goto package switch
const fallthrough if range type
continue for import return var

We've seen some of these already. We used var and const in the section, and we briefly looked at package and import in our "Hello World" program at the start of the chapter. Others need more attention and have their own chapter or section:

Control Structures

There are only a few control structures in Go. To write loops we use the for keyword, and there is a switch and of course an if. When working with channels select will be used (see ). Parentheses are are not required around the condition, and the body must always be brace-delimited.

If-Else

In Go an if looks like this:

if x > 0 {
    return y
} else {
    return x
}

Since if and switch accept an initialization statement, it's common to see one used to set up a (local) variable.

if err := SomeFunction(); err == nil {
    // do something
} else {
    return err
}

It is idomatic in Go to omit the else when the if statement's body has a break, continue, return or, goto, so the above code would be better written as:

if err := SomeFunction(); err != nil {
    return err
}
// do something

The opening brace on the first line must be positioned on the same line as the if statement. There is no arguing about this, because this is what gofmt outputs.

Goto

Go has a goto statement - use it wisely. With goto you jump to a label which must be defined within the current function. For instance, a loop in disguise:

func myfunc() {
    i := 0
Here:
    fmt.Println(i)
    i++
    goto Here
}

The string Here: indicates a label. A label does not need to start with a capital letter and is case sensitive.

For

The Go for loop has three forms, only one of which has semicolons:

Short declarations make it easy to declare the index variable right in the loop.

sum := 0
for i := 0; i < 10; i++ {
    sum = sum + i
}

Note that the variable i ceases to exist after the loop.

Break and Continue

With break you can quit loops early. By itself, break breaks the current loop.

for i := 0; i < 10; i++ {
    if i > 5 {
    break 1
    }
    fmt.Println(i) 2
}

Here we break the current loop 1, and don't continue with the fmt.Println(i) statement 2. So we only print 0 to 5. With loops within loop you can specify a label after break to identify which loop to stop:

J:  for j := 0; j < 5; j++ { 1
        for i := 0; i < 10; i++ {
            if i > 5 {
                break J 2
            }
            fmt.Println(i)
        }
    }

Here we define a label "J" 1, preceding the for-loop there. When we use break J 2, we don't break the inner loop but the "J" loop.

With continue you begin the next iteration of the loop, skipping any remaining code. In the same way as break, continue also accepts a label.

Range

The keyword range can be used for loops. It can loop over slices, arrays, strings, maps and channels (see ). range is an iterator that, when called, returns the next key-value pair from the "thing" it loops over. Depending on what that is, range returns different things.

When looping over a slice or array, range returns the index in the slice as the key and value belonging to that index. Consider this code:

list := []string{"a", "b", "c", "d", "e", "f"}
for k, v := range list {
    // do something with k and v
}

First we create a slice of strings. Then we use range to loop over them. With each iteration, range will return the index as an int and the key as a string. It will start with 0 and "a", so k will be 0 through 5, and v will be "a" through "f".

You can also use range on strings directly. Then it will break out the individual Unicode characters 6 Mostly, when people talk about characters, they mean 8 bit characters. As UTF-8 characters may be up to 32 bits the word rune is used. In this case the type of char is rune. and their start position, by parsing the UTF-8. The loop:

for pos, char := range "Gő!" {
    fmt.Printf("character '%c' starts at byte position %d\n", char, pos)
}

prints

character 'G' starts at byte position 0
character 'ő' starts at byte position 1
character '!' starts at byte position 3

Note that ő took 2 bytes, so '!' starts at byte 3.

Switch

Go's switch is very flexible; you can match on much more than just integers. The cases are evaluated top to bottom until a match is found, and if the switch has no expression it switches on true. It's therefore possible -- and idiomatic -- to write an if-else-if-else chain as a switch.

 Convert hexadecimal character to an int value
switch { 1
case '0' <= c && c <= '9': 2
    return c - '0' 3
case 'a' <= c && c <= 'f': 4
    return c - 'a' + 10
case 'A' <= c && c <= 'F': 5
    return c - 'A' + 10
}
return 0

A switch without a condition is the same as switch true 1. We list the different cases. Each case statement has a condition that is either true of false. Here 2 we check if c is a number. If c is a number we return its value 3. Check if c falls between "a" and "f" 4. For an "a" we return 10, for "b" we return 11, etc. We also do the same 5 thing for "A" to "F".

There is no automatic fall through, you can use fallthrough for that.

switch i {
    case 0:  fallthrough
    case 1: 1
        f()
    default:
        g() 2

f() can be called when i == 0 1. With default you can specify an action when none of the other cases match. Here g() is called when i is not 0 or 1 2. We could rewrite the above example as:

switch i {
    case 0, 1: 1
        f()
    default:
        g()

You can list cases on one line 1, separated by commas.

Built-in Functions

A few functions are predefined, meaning you don't have to include any package to get access to them. lists them all.7

Pre-defined functions in Go.
       
close new panic complex
delete make recover real
len append print imag
cap copy println

These built-in functions are documented in the builtin pseudo package that is included in recent Go releases. Let's go over these functions briefly.

close
is used in channel communication. It closes a channel. We'll learn more about this in .
delete
is used for deleting entries in maps.
len and cap
are used on a number of different types, len is used to return the lengths of strings, slices, and arrays. In the next section we'll look at slices, arrays and the function cap.
new
is used for allocating memory for user defined data types. See .
make
is used for allocating memory for built-in types (maps, slices, and channels). See .
copy, append
copy is for copying slices. And append is for concatenating slices. See in this chapter.
panic, recover
are used for an exception mechanism. See for more.
print, println
are low level printing functions that can be used without reverting to the fmt package. These are mainly used for debugging.
complex, real, imag
all deal with complex numbers. We will not use complex numbers in this book.

Arrays, Slices, and Maps

To store multiple values in a list, you can use arrays, or their more flexible cousin: slices. A dictionary or hash type is also available. It is called a map in Go.

Arrays

An array is defined by: [n]<type>, where \(n\) is the length of the array and <type> is the stuff you want to store. To assign or index an element in the array, you use square brackets:

var arr [10]int
arr[0] = 42
arr[1] = 13
fmt.Printf("The first element is %d\n", arr[0])

Array types like var arr [10]int have a fixed size. The size is part of the type. They can't grow, because then they would have a different type. Also arrays are values: Assigning one array to another copies all the elements. In particular, if you pass an array to a function it will receive a copy of the array, not a pointer to it.

To declare an array you can use the following: var a [3]int. To initialize it to something other than zero, use a composite literal : a := [3]int{1, 2, 3}. This can be shortened to a := [...]int{1, 2, 3}, where Go counts the elements automatically.

When declaring arrays you always have to type something in between the square brackets, either a number or three dots (...), when using a composite literal. When using multidimensional arrays, you can use the following syntax: a := [2][2]int{ {1,2}, {3,4} }. Now that you know about arrays you will be delighted to learn that you will almost never use them in Go, because there is something much more flexible: slices.

Slices

A slice is similar to an array, but it can grow when new elements are added. A slice always refers to an underlying array. What makes slices different from arrays is that a slice is a pointer to an array; slices are reference types.

That means that if you assign one slice to another, both refer to the same underlying array. For instance, if a function takes a slice argument, changes it makes to the elements of the slice will be visible to the caller, analogous to passing a pointer to the underlying array. With: slice := make([]int, 10), you create a slice which can hold ten elements. Note that the underlying array isn't specified. A slice is always coupled to an array that has a fixed size. For slices we define a capacity and a length . shows the creation of an array, then the creation of a slice. First we create an array of \(m\) elements of the type int: var array[m]int .

Next, we create a slice from this array: slice := array[:n] . And now we have:

Array versus slice
An array versus a slice.

Given an array, or another slice, a new slice is created via a[n:m]. This creates a new slice which refers to the variable a, starts at index n, and ends before index m. It has length n - m.

a := [...]int{1, 2, 3, 4, 5} 1
s1 := a[2:4] 2
s2 := a[1:5] 3
s3 := a[:]   4
s4 := a[:4]  5
s5 := s2[:] 6
s6 := a[2:4:5] 7

First we define 1 an array with five elements, from index 0 to 4. From this we create 2 a slice with the elements from index 2 to 3, this slices contains: 3, 4. Then we we create another slice 3 from a: with the elements from index 1 to 4, this contains: 2, 3, 4, 5. With a[:] 4 we create a slice with all the elements in the array. This is a shorthand for: a[0:len(a)]. And with a[:4] 5 we create a slice with the elements from index 0 to 3, this is short for: a[0:4], and gives us a slices that contains: 1, 2, 3, 4. With s2[:] we create a slice from the slice s2 6, note that s5 still refers to the array a. Finally, we create a slice with the elements from index 3 to 3 and also set the cap to 4 7.

When working with slices you can overrun the bounds, consider this code.

package main

func main() {
	var array [100]int   1
	slice := array[0:99] 2

	slice[98] = 1 3
	slice[99] = 2 4
}

At 1 we create an array with a 100 elements, indexed from 0 to 99. Then at 2 we create a slice that has index 0 to 98. We assign 1 to the 99th element 3 of the slice. This works as expected. But at 4 we dare to do the impossible, and and try to allocate something beyond the length of the slice and we are greeted with a runtime error: Error: "throw: index out of range".

If you want to extend a slice, there are a couple of built-in functions that make life easier: append and copy. The append function appends zero or more values to a slice and returns the result: a slice with the same type as the original. If the original slice isn't big enough to fit the added values, append will allocate a new slice that is big enough. So the slice returned by append may refer to a different underlying array than the original slice does. Here's an example:

s0 := []int{0, 0}
s1 := append(s0, 2) 1
s2 := append(s1, 3, 5, 7) 2
s3 := append(s2, s0...) 3

At 1 we append a single element, making s1 equal to []int{0, 0, 2}. At 2 we append multiple elements, making s2 equal to []int{0, 0, 2, 3, 5, 7}. And at 3 we append a slice, giving us s3 equal to []int{0, 0, 2, 3, 5, 7, 0, 0}. Note the three dots used after s0...! This is needed make it clear explicit that you're appending another slice, instead of a single value.

The copy function copies slice elements from a source to a destination, and returns the number of elements it copied. This number is the minimum of the length of the source and the length of the destination. For example:

var a = [...]int{0, 1, 2, 3, 4, 5, 6, 7}
var s = make([]int, 6)
n1 := copy(s, a[0:])  1
n2 := copy(s, s[2:])  2

After 1, n1 is 6, and s is []int{0, 1, 2, 3, 4, 5}. And after 2, n2 is 4, and s is []int{2, 3, 4, 5, 4, 5}.

Maps

Many other languages have a type similar to maps built-in. For instance, Perl has hashes, Python has its dictionaries, and C++ also has maps (as part of the libraries). In Go we have the map type. A map can be thought of as an array indexed by strings (in its most simple form).

monthdays := map[string]int{
    "Jan": 31, "Feb": 28, "Mar": 31,
    "Apr": 30, "May": 31, "Jun": 30,
    "Jul": 31, "Aug": 31, "Sep": 30,
    "Oct": 31, "Nov": 30, "Dec": 31, 1
}

The general syntax for defining a map is map[<from type>]<to type>. Here, we define a map that converts from a string (month abbreviation) to an int (number of days in that month). Note that the trailing comma at 1 is required.

Use make when only declaring a map: monthdays := make(map[string]int). A map is a reference type.

For indexing ("searching") the map, we use square brackets. For example, suppose we want to print the number of days in December: fmt.Printf("%d\n", monthdays["Dec"])

If you are looping over an array, slice, string, or map a, range clause will help you again, it returns the key and corresponding value with each invocation.

year := 0
for _, days := range monthdays 1
    year += days
}
fmt.Printf("Numbers of days in a year: %d\n", year)

At 1 we use the underscore to ignore (assign to nothing) the key returned by range. We are only interested in the values from monthdays.

To add elements to the map, you would add new month with: monthdays["Undecim"] = 30. If you use a key that already exists, the value will be silently overwritten: monthdays["Feb"] = 29. To test for existence , you would use the following: value, present := monthdays["Jan"]. If the key "Jan" exists, present will be true. It's more Go like to name present "ok", and use: v, ok := monthdays["Jan"]. In Go we call this the "comma ok" form.

You can remove elements from the map: delete(monthdays, "Mar") 8. In general the syntax delete(m, x) will delete the map entry retrieved by the expression m[x].

Exercises

For-loop

  1. Create a loop with the for construct. Make it loop 10 times and print out the loop counter with the fmt package.

  2. Rewrite the loop from 1 to use goto. The keyword for may not be used.

  3. Rewrite the loop again so that it fills an array and then prints that array to the screen.

Answer

  1. There are many possibilities. One solutions could be:

    package main
    
    import "fmt"
    
    func main() {
    	for i := 0; i < 10; i++ {
    		fmt.Println("%d", i)
    	}
    }
    
    Let's compile this and look at the output.

    % go build for.go
    % ./for
    0
    1
    .
    .
    .
    9
    
  2. Rewriting the loop results in code that should look something like this (only showing the main-function):

    func main() {
        i := 0  1
    Loop:           2
        if i < 10 {
            fmt.Printf("%d\n", i)
            i++
            goto Loop 3
        }
    }
    

    At 1 we define our loop variable. And at 2 we define a label and at 3 we jump to this label.

  3. The following is one possible solution:

    package main
    
    import "fmt"
    
    func main() {
    	var arr [10]int 1
    	for i := 0; i < 10; i++ {
    		arr[i] = i 2
    	}
    	fmt.Printf("%v", arr) 3
    }
    

    Here 1 we create an array with 10 elements. Which we then fill 2 one by one. And finally we print it 3 with %v which lets Go to print the value for us. You could even do this in one fell swoop by using a composite literal:

fmt.Printf("%v\n", [...]int{0,1,2,3,4,5,6,7,8,9})

Average

  1. Write code to calculate the average of a float64 slice. In a later exercise you will make it into a function.

Answer

  1. The following code calculates the average.
sum := 0.0
switch len(xs) {
case 0: 1
    avg = 0
default: 2
    for _, v := range xs {
        sum += v
    }
    avg = sum / float64(len(xs)) 3
}

Here at 1 we check if the length is zero and if so, we return 0. Otherwise we calculate the average at 2. We have to convert the value return from len to a float64 to make the division work at 3.

FizzBuzz

  1. Solve this problem, called the Fizz-Buzz problem:

Write a program that prints the numbers from 1 to 100. But for multiples of three print, "Fizz" instead of the number, and for multiples of five, print "Buzz". For numbers which are multiples of both three and five, print "FizzBuzz".

Answer

  1. A possible solution to this problem is the following program.

package main

import "fmt"

func main() {
	const (
		FIZZ = 3 1
		BUZZ = 5
	)
	var p bool                 2
	for i := 1; i < 100; i++ { 3
		p = false
		if i%FIZZ == 0 { 4
			fmt.Printf("Fizz")
			p = true
		}
		if i%BUZZ == 0 { 5
			fmt.Printf("Buzz")
			p = true
		}
		if !p { 6
			fmt.Printf("%v", i)
		}
		fmt.Println()
	}
}

Here 1 we define two constants to make our code more readable, see . At 2 we define a boolean that keeps track if we already printed something. At 3 we start our for-loop, see . If the value is divisible by FIZZ - that is, 3 - , we print "Fizz" 4. And at 5 we check if the value is divisble by BUZZ -- that is, 5 -- if so print "Buzz". Note that we have also taken care of the FizzBuzz case. At 6, if printed neither Fizz nor Buzz printed, we print the value.

Functions

I'm always delighted by the light touch and stillness of early programming languages. Not much text; a lot gets done. Old programs read like quiet conversations between a well-spoken research worker and a well- studied mechanical colleague, not as a debate with a compiler. Who'd have guessed sophistication bought such noise?

Functions are the basic building blocks of Go programs; all interesting stuff happens in them.

Here is an example of how you can declare a function:

type mytype int
func (p mytype) funcname(q int) (r,s int) { return 0,0 }
 1        2        3      4        5         6

To declare a function, you use the func keyword 1. You can optionally bind 2 to a specific type called receiver (a function with a receiver is usually called a method). This will be explored in . Next 3 you write the name of your function. Here 4 we define that the variable q of type int is the input parameter. Parameters are passed pass-by-value. The variables r and s 5 are the named return parameters for this function. Functions in Go can have multiple return values. This is very useful to return a value and error. This removes the need for in-band error returns (such as -1 for EOF) and modifying an argument. If you want the return parameters not to be named you only give the types: (int, int). If you have only one value to return you may omit the parentheses. If your function is a subroutine and does not have anything to return you may omit this entirely. Finally, we have the body 6 of the function. Note that return is a statement so the braces around the parameter(s) are optional.

As said the return or result parameters of a Go function can be given names and used as regular variables, just like the incoming parameters. When named, they are initialized to the zero values for their types when the function begins. If the function executes a return statement with no arguments, the current values of the result parameters are returned. Using these features enables you (again) to do more with less code.9

The names are not mandatory but they can make code shorter and clearer: they are documentation. However don't overuse this feature, especially in longer functions where it might not be immediately apparent what is returned.

Functions can be declared in any order you wish. The compiler scans the entire file before execution, so function prototyping is a thing of the past in Go. Go does not allow nested functions, but you can work around this with anonymous functions. See the Section in this chapter. Recursive functions work just as in other languages:

func rec(i int) {
    if i == 10 { 1
        return
    }
    rec(i+1) 2
    fmt.Printf("%d ", i)
}

Here 2 we call the same function again, rec returns when i has the value 10, this is checked on the second line 1. This function prints: 9 8 7 6 5 4 3 2 1 0, when called as rec(0).

Scope

Variables declared outside any functions are global in Go, those defined in functions are local to those functions. If names overlap - a local variable is declared with the same name as a global one - the local variable hides the global one when the current function is executed.

In the following example we call g() from f():

package main

var a int  1

func main() {
    a = 5
    print(a)
    f()
}

func f() {
    a := 6  2
    print(a)
    g()
}

func g() {
    print(a)
}

Here 1, we declare a to be a global variable of type int. Then in the main function we give the global a the value of 5, after printing it we call the function f. Then here 2, a := 6, we create a new, local variable also called a. This new a gets the value of 6, which we then print. Then we call g, which uses the global a again and prints a's value set in main. Thus the output will be: 565. A local variable is only valid when we are executing the function in which it is defined. Note that the := used in line 12 is sometimes hard to spot so it is generally advised not to use the same name for global and local variables.

Functions as values

As with almost everything in Go, functions are also just values. They can be assigned to variables as follows:

import "fmt"

func main() {
	a := func() { 1
		fmt.Println("Hello")
	} 2
	a() 3
}

a is defined as an anonymous (nameless) function 1. Note the lack of parentheses () after a. If there were, that would be to call some function with the name a before we have defined what a is. Once a is defined, then we can call it, 3.

Functions--as--values may be used in other places, for example maps. Here we convert from integers to functions:

var xs = map[int]func() int{
    1: func() int { return 10 },
    2: func() int { return 20 },
    3: func() int { return 30 },
}

Note that the final comma on second to last line is mandatory.

Or you can write a function that takes a function as its parameter, for example a Map function that works on int slices. This is left as an exercise for the reader; see the exercise .

Callbacks

Because functions are values they are easy to pass to functions, from where they can be used as callbacks. First define a function that does "something" with an integer value:

func printit(x int) {
    fmt.Printf("%v\n", x)
}

This function does not return a value and just prints its argument. The signature of this function is: func printit(int), or without the function name: func(int). To create a new function that uses this one as a callback we need to use this signature:

func callback(y int, f func(int)) {
    f(y)
}

Here we create a new function that takes two parameters: y int, i.e. just an int and f func(int), i.e. a function that takes an int and returns nothing. The parameter f is the variable holding that function. It can be used as any other function, and we execute the function on line 2 with the parameter y: f(y)

Deferred Code

Suppose you have a function in which you open a file and perform various writes and reads on it. In such a function there are often spots where you want to return early. If you do that, you will need to close the file descriptor you are working on. This often leads to the following code:

func ReadWrite() bool {
    file.Open("file")
     Do your thing
    if failureX {
        file.Close() 1
        return false
    }

    if failureY {
        file.Close() 2
        return false
    }
    file.Close() 3
    return true  4
}

Note that we repeat a lot of code here; you can see the that file.Close() is called at 1 2 3. To overcome this, Go has the defer keyword. After defer you specify a function which is called just before 4 the current function exits.

With defer we can rewrite the above code as follows. It makes the function more readable and it puts the Close right next to the Open.

func ReadWrite() bool {
    file.Open("filename")
    defer file.Close() 1
     Do your thing
    if failureX {
        return false 2
    }
    if failureY {
        return false 3
    }
    return true 4
}

At 1 file.Close() is added to the defer list. Close is now done automatically at 2 3 4. This makes the function shorter and more readable. It puts the Close right next to the Open.

You can put multiple functions on the "defer list", like this example from

for i := 0; i < 5; i++ {
    defer fmt.Printf("%d ", i)
}

Deferred functions are executed in LIFO order, so the above code prints: 4 3 2 1 0.

With defer you can even change return values, provided that you are using named result parameters and a function literal 10, i.e:

defer func() {/* ... */}()

Here we use a function without a name and specify the body of the function inline, basically we're creating a nameless function on the spot. The final braces are needed because defer needs a function call, not a function value. If our anonymous function would take an parameter it would be easier to see why we need the braces:

defer func(x int) {/* ... */}(5)

In this (unnamed) function you can access any named return parameter:

func f() (ret int)
    defer func() { 1
        ret++
    }()
    return 0
}

Here 1 we specify our function, the named return value ret is initialized with zero. The nameless function in the defer increments the value of ret with 1. The return 0 on line 5 will not be the returned value, because of defer. The function f will return 1!

Variadic Parameter

Functions that take a variable number of parameters are known as variadic functions. To declare a function as variadic, do something like this:

func myfunc(arg ...int) {}

The arg ...int instructs Go to see this as a function that takes a variable number of arguments. Note that these arguments all have to have the type int. In the body of your function the variable arg is a slice of ints:

for _, n := range arg {
    fmt.Printf("And the number is: %d\n", n)
}

We range over the arguments on the first line. We are not interested in the index as returned by range, hence the use of the underscore there. In the body of the range we just print the parameters we were given.

If you don't specify the type of the variadic argument it defaults to the empty interface interface{} (see Chapter ).

Suppose we have another variadic function called myfunc2, the following example shows how to pass variadic arguments to it:

func myfunc(arg ...int) {
    myfunc2(arg...)
    myfunc2(arg[:2]...)
}

With myfunc2(arg...) we pass all the parameters to myfunc2, but because the variadic parameters is just a slice, we can use some slice tricks as well.

Panic and recovering

Go does not have an exception mechanism: you cannot throw exceptions. Instead it uses a panic-and-recover mechanism. It is worth remembering that you should use this as a last resort, your code will not look, or be, better if it is littered with panics. It's a powerful tool: use it wisely. So, how do you use it? In the words of the Go Authors :

Panic
is a built-in function that stops the ordinary flow of control and begins panicking. When the function F calls panic, execution of F stops, any deferred functions in F are executed normally, and then F returns to its caller. To the caller, F then behaves like a call to panic. The process continues up the stack until all functions in the current goroutine have returned, at which point the program crashes. Panics can be initiated by invoking panic directly. They can also be caused by runtime errors, such as out-of-bounds array accesses.
Recover
is a built-in function that regains control of a panicking goroutine. Recover is only useful inside deferred functions. During normal execution, a call to recover will return nil and have no other effect. If the current goroutine is panicking, a call to recover will capture the value given to panic and resume normal execution.

This function checks if the function it gets as argument will panic when it is executed11:

func Panic(f func()) (b bool) { 1
    defer func() { 2
        if x := recover(); x != nil {
            b = true
        }
    }()
    f() 3
    return 4
}

We define a new function Panic 1 that takes a function as an argument (see ). It returns true if f panics when run, else false. We then 2 define a defer function that utilizes recover. If the current goroutine panics, this defer function will notice that. If recover() returns non-nil we set b to true. At 3 Execute the function we received as the argument. And finally 4 we return the value of b. Because b is a named return parameter.

The following code fragment, shows how we can use this function:

func panicy() {
    var a []int
    a[3] = 5
}

func main() {
    fmt.Println(Panic(panicy))
}

On line 3 the a[3] = 5 triggers a runtime out of bounds error which results in a panic. Thus this program will print true. If we change line 2: var a []int to var a [4]int the function panicy does not panic anymore. Why?

Exercises

Average

  1. Write a function that calculates the average of a float64 slice.

Answer

  1. The following function calculates the average:

package main

func average(xs []float64) (avg float64) { 1
	sum := 0.0
	switch len(xs) {
	case 0:                 2
		avg = 0
	default:                3
		for _, v := range xs {
			sum += v
		}
		avg = sum / float64(len(xs)) 4
	}
	return  5
}

At 1 we use a named return parameter. If the length of xs is zero 2, we return 0. Otherwise 3, we calculate the average. At 4 we convert the value to a float64 to make the division work as len returns an int. Finally, at 5 we reutrn our avarage.

Bubble sort

  1. Write a function that performs a bubble sort on a slice of ints. From :

It works by repeatedly stepping through the list to be sorted, comparing each pair of adjacent items and swapping them if they are in the wrong order. The pass through the list is repeated until no swaps are needed, which indicates that the list is sorted. The algorithm gets its name from the way smaller elements "bubble" to the top of the list.

It also gives an example in pseudo code:

procedure bubbleSort( A : list of sortable items )
  do
    swapped = false
    for each i in 1 to length(A) - 1 inclusive do:
      if A[i-1] > A[i] then
        swap( A[i-1], A[i] )
        swapped = true
      end if
    end for
  while swapped
end procedure

Answer

  1. Bubble sort isn't terribly efficient. For \(n\) elements it scales \(O(n^2)\). But bubble sort is easy to implement:

    
    func main() {
    	n := []int{5, -1, 0, 12, 3, 5}
    	fmt.Printf("unsorted %v\n", n)
    	bubblesort(n)
    	fmt.Printf("sorted %v\n", n)
    }
    
    func bubblesort(n []int) {
    	for i := 0; i < len(n)-1; i++ {
    		for j := i + 1; j < len(n); j++ {
    			if n[j] < n[i] {
    				n[i], n[j] = n[j], n[i]
    			}
    		}
    

    Because a slice is a reference type, the bubblesort function works and does not need to return a sorted slice.

For-loop II

  1. Take what you did in exercise to write the for loop and extend it a bit. Put the body of the for loop - the fmt.Printf - in a separate function.

Answer

1.

package main

import "fmt"

func main() {
	for i := 0; i < 10; i++ {
		show(i)
	}
}

func show(j int) {
	fmt.Printf("%d\n", j)
}

Fibonacci

  1. The Fibonacci sequence starts as follows: \(1, 1, 2, 3, 5, 8, 13, \ldots\) Or in mathematical terms: \( x_1 = 1; x_2 = 1; x_n = x_{n-1} + x_{n-2}\quad\forall n > 2 \).

    Write a function that takes an int value and gives that many terms of the Fibonacci sequence.

Answer

  1. The following program calculates Fibonacci numbers:

package main

import "fmt"

func fibonacci(value int) []int {
	x := make([]int, value) 1
	x[0], x[1] = 1, 1       2
	for n := 2; n < value; n++ {
		x[n] = x[n-1] + x[n-2] 3
	}
	return x 4
}

func main() {
	for _, term := range fibonacci(10) { 5
		fmt.Printf("%v ", term)
	}
}

At 1 we create an array to hold the integers up to the value given in the function call. At 2 we start the Fibonacci calculation. Then 3: \(x_n = x_{n-1} + x_{n-2}\). At 4 we return the entire array. And at 5 we use the range keyword to "walk" the numbers returned by the Fibonacci function. Here up to 10. Finally, we print the numbers.

Var args

  1. Write a function that takes a variable number of ints and print each integer on a separate line.

Answer

  1. For this we need the {...}-syntax to signal we define a function that takes an arbitrary number of arguments.

    package main
    
    import "fmt"
    
    func main() {
    	printthem(1, 4, 5, 7, 4)
    	printthem(1, 2, 4)
    }
    
    func printthem(numbers ...int) {
    	for _, d := range numbers {
    		fmt.Printf("%d\n", d)
    	}
    }
    

Functions that return functions

  1. Write a function that returns a function that performs a \(+2\) on integers. Name the function plusTwo. You should then be able do the following:

    p := plusTwo()
    fmt.Printf("%v\n", p(2))
    

    Which should print 4. See .

  2. Generalize the function from above and create a plusX(x) which returns functions that add x to an integer.

Answer

  1. Define a new function that returns a function: return func(x int) int { return x + 2 } Function literals at work, we define the +2--function right there in the return statement.

    func main() {
       p2 := plusTwo()
       fmt.Printf("%v\n",p2(2))
    }
    
    func plusTwo() func(int) int { 1
        return func(x int) int { return x + 2 } 2
    }
    
  2. Here we use a closure:

    func plusX(x int) func(int) int { 1
       return func(y int) int { return x + y } 2
    }
    

    Here 1, we again define a function that returns a function. We use the local variable x in the function literal at 2.

Maximum

  1. Write a function that finds the maximum value in an int slice ([]int).

Answer

  1. This function returns the largest int in the slice \var{l}:

    func max(l []int) (max int) {   1
        max = l[0]
        for _, v := range l {   2
            if v > max {    3
                max = v
            }
        }
        return 4
    }
    

    At 1 we use a named return parameter. At 2 we loop over l. The index of the element is not important. At 3, if we find a new maximum, we remember it. And at 4 we have a "lone" return; the current value of max is now returned.

Map function

A map()-function is a function that takes a function and a list. The function is applied to each member in the list and a new list containing these calculated values is returned. Thus:

\[ \mathrm{map}(f(), (a_1,a_2,\ldots,a_{n-1},a_n)) = (f(a_1), f(a_2),\ldots,f(a_{n-1}), f(a_n)) \]

  1. Write a simple map()-function in Go. It is sufficient for this function only to work for ints.

Answer

  1. A possible answer:

    func Map(f func(int) int, l []int) []int {
        j := make([]int, len(l))
        for k, v := range l {
            j[k] = f(v)
        }
        return j
    }
    
    func main() {
        m := []int{1, 3, 4}
        f := func(i int) int {
            return i * i
        }
        fmt.Printf("%v", (Map(f, m)))
    }
    

Stack

  1. Create a simple stack which can hold a fixed number of ints. It does not have to grow beyond this limit. Define push -- put something on the stack -- and pop -- retrieve something from the stack -- functions. The stack should be a LIFO (last in, first out) stack.

A stack
A stack.

  1. Write a String method which converts the stack to a string representation. The stack in the figure could be represented as: [0:m] [1:l] [2:k] .

Answer

  1. First we define a new type that represents a stack; we need an array (to hold the keys) and an index, which points to the last element. Our small stack can only hold 10 elements.

    type stack struct {
        i    int
        data [10]int
    }
    

Next we need the push and pop functions to actually use the thing. First we show the wrong solution!

In Go, data passed to functions is passed-by-value meaning a copy is created and given to the function. The first stab for the function push could be:

func (s stack) push(k int) {
    if s.i+1 > 9 {
            return
    }
    s.data[s.i] = k
    s.i++
}

The function works on the s which is of the type stack. To use this we just call s.push(50), to push the integer 50 on the stack. But the push function gets a copy of s, so it is not working on the real thing. Nothing gets pushed to our stack. For example the following code:

var s stack
s.push(25)
fmt.Printf("stack %v\n", s);
s.push(14)
fmt.Printf("stack %v\n", s);

prints:

stack [0:0]
stack [0:0]

To solve this we need to give the function push a pointer to the stack. This means we need to change push from

func (s stack) push(k int)

to

func (s *stack) push(k int).

We should now use new() (see ). in to create a pointer to a newly allocated stack, so line 1 from the example above needs to be s := new(stack) .

And our two functions become:

func (s *stack) push(k int) {
    s.data[s.i] = k
    s.i++
}

func (s *stack) pop() int {
    s.i--
    ret := s.data[s.i]
    s.data[s.i] = 0
    return ret
}

Which we then use as follows:

func main() {
    var s stack
    s.push(25)
    s.push(14)
    fmt.Printf("stack %v\n", s)
}
  1. fmt.Printf("%v") can print any value (%v) that satisfies the Stringer interface (see ). For this to work we only need to define a String() function for our type:

    func (s stack) String() string {
        var str string
        for i := 0; i <= s.i; i++ {
            str = str + "[" +
                strconv.Itoa(i) + ":" + strconv.Itoa(s.data[i]) + "]"
        }
        return str
    }
    

Packages

"^"

A package is a collection of functions and data.

You declare a package with the package keyword. The filename does not have to match the package name. The convention for package names is to use lowercase characters. Go packages may consist of multiple files, but they share the package <name> line. Let's define a package even in the file even.go.

package even 1

func Even(i int) bool {	2
	return i % 2 == 0
}

func odd(i int) bool {	3
	return i % 2 == 1
}

Here 1 we start a new namespace: "even". The function Even 2 starts with a capital letter. This means the function is exported, and may be used outside our package (more on that later). The function odd 3 does not start with a capital letter, so it is a private function.

Now we just need to build the package. We create a directory under $GOPATH, and copy even.go there (see in ).

% mkdir $GOPATH/src/even
% cp even.go $GOPATH/src/even
% go build
% go install

Now we can use the package in our own program myeven.go:

package main

import ( 1
	"even"	2
	"fmt"	3
)

func main() {
	i := 5
	fmt.Printf("Is %d even? %v\n", i, even.Even(i)) 4
}

Import 1 the following packages. The local package even is imported here 2. This 3 imports the official fmt package. And now we use 4 the function from the even package. The syntax for accessing a function from a package is <package>.FunctionName(). And finally we can build our program.

% go build myeven.go
% ./myeven
Is 5 even? false

If we change our myeven.go at <4> to use the unexported function even.odd: fmt.Printf("Is %d even? %v\n", i, even.odd(i)) We get an error when compiling, because we are trying to use a private function:

myeven.go: cannot refer to unexported name even.odd

Note that the "starts with capital \(\rightarrow\) exported", "starts with lower-case \(\rightarrow\) private" rule also extends to other names (new types, global variables) defined in the package. Note that the term "capital" is not limited to US-ASCII -- it extends to all bicameral alphabets (Latin, Greek, Cyrillic, Armenian and Coptic).

Identifiers

The Go standard library names some function with the old (Unix) names while others are in CamelCase. The convention is to leave well-known legacy not-quite-words alone rather than try to figure out where the capital letters go: Atoi, Getwd, Chmod. CamelCasing works best when you have whole words to work with: ReadFile, NewWriter, MakeSlice. The convention in Go is to use CamelCase rather than underscores to write multi-word names.

As we did above in our myeven program, accessing content from an imported (with import ) package is done with using the package's name and then a dot. After import "bytes" the importing program can talk about bytes.Buffer. A package name should be good, short, concise and evocative. The convention in Go is that package names are lowercase, single word names.

The package name used in the import statement is the default name used. But if the need arises (two different packages with the same name for instance), you can override this default: import bar "bytes" The function Buffer is now accessed as bar.Buffer.

Another convention is that the package name is the base name of its source directory; the package in src/compress/gzip is imported as compress/gzip but has name gzip, not compress/gzip.

It is important to avoid stuttering when naming things. For instance, the buffered reader type in the bufio package is called Reader, not BufReader, because users see it as bufio.Reader, which is a clear, concise name.

Similarly, the function to make new instances of ring.Ring (package container/ring), would normally be called NewRing, but since Ring is the only type exported by the package, and since the package is called ring, it's called just New. Clients of the package see that as ring.New. Use the package structure to help you choose good names.

Another short example is once.Do (see package sync); once.Do(setup) reads well and would not be improved by writing once.DoOrWaitUntilDone(setup). Long names don't automatically make things more readable.

Documenting packages

When we created our even package, we skipped over an important item: documentation. Each package should have a package comment, a block comment preceding the package clause. In our case we should extend the beginning of the package, with:

// The even package implements a fast function for detecting if an integer
// is even or not.
package even

When running go doc this will show up at the top of the page. When a package consists of multiple files the package comment should only appear in one file. A common convention (in really big packages) is to have a separate doc.go that only holds the package comment. Here is a snippet from the official regexp package:

/*
    The regexp package implements a simple library for
    regular expressions.

    The syntax of the regular expressions accepted is:

    regexp:
        concatenation { '|' concatenation }
*/
package regexp

Each defined (and exported) function should have a small line of text documenting the behavior of the function. Again to extend our even package:

// Even returns true of i is even. Otherwise false is returned.
func Even(i int) bool {

And even though odd is not exported, it's good form to document it as well.

// odd is the opposite of Even.
func odd(i int) bool {

Testing packages

In Go it is customary to write (unit) tests for your package. Writing tests involves the testing package and the program go test. Both have excellent documentation.

The go test program runs all the test functions. Without any defined tests for our even package, go test yields:

% go test
?       even    [no test files]

Let us fix this by defining a test in a test file. Test files reside in the package directory and are named *_test.go. Those test files are just like other Go programs, but go test will only execute the test functions. Each test function has the same signature and its name should start with Test: func TestXxx(t *testing.T) .

When writing test you will need to tell go test whether a test was successful or not. A successful test function just returns. When the test fails you can signal this with the following functions. These are the most important ones (see go doc testing or go help testfunc for more):

Putting all this together we can write our test. First we pick a name: even_test.go. Then we add the following contents:

package even 1

import "testing" 2

func TestEven(t *testing.T) { 3
	if !Even(2) {
		t.Log("2 should be even!")
		t.Fail()
	}
}

A test file belongs to the current 1 package. This is not only convenient, but also allows tests of unexported functions and structures. We then 2 import the testing package. And finally the test we want to execute. The code here 3 should hold no surprises: we check if the Even function works OK. And now, the moment we have been waiting form executing the test.

% go test
ok      even    0.001s

Our test ran and reported ok. Success! If we redefine our test function, we can see the result of a failed test:

// Entering the twilight zone
func TestEven(t *testing.T) {
    if Even(2) {
        t.Log("2 should be odd!")
        t.Fail()
    }
}

We now get:

FAIL    even    0.004s
--- FAIL: TestEven (0.00 seconds)
    2 should be odd!
FAIL

And you can act accordingly (by fixing the test for instance).

Writing new packages should go hand in hand with writing (some) documentation and test functions. It will make your code better and it shows that you really put in the effort.

The Go test suite also allows you to incorporate example functions which serve as documentation and as tests. These functions need to start with Example.

func ExampleEven() {
    if Even(2) {
        fmt.Printf("Is even\n")
    }
     Output: 1
     Is even
}

Those last two comments lines 1 are part of the example, go test uses those to check the generated output with the text in the comments. If there is a mismatch the test fails.

Useful packages

The standard libary of Go includes a huge number of packages. It is very enlightening to browse the $GOROOT/src/pkg directory and look at the packages. We cannot comment on each package, but the following are worth a mention: 12

fmt

Package fmt implements formatted I/O with functions analogous to C's printf and scanf. The format verbs are derived from C's but are simpler. Some verbs (%-sequences) that can be used:

%v

The value in a default format. when printing structs, the plus flag (%+v) adds field names.
%#v
a Go-syntax representation of the value.
%T
a Go-syntax representation of the type of the value.
io

This package provides basic interfaces to I/O primitives. Its primary job is to wrap existing implementations of such primitives, such as those in package os, into shared public interfaces that abstract the functionality, plus some other related primitives.

bufio

This package implements buffered I/O. It wraps an io.Reader or io.Writer object, creating another object (Reader or Writer) that also implements the interface but provides buffering and some help for textual I/O.

sort

The sort package provides primitives for sorting arrays and user-defined collections.

strconv

The strconv package implements conversions to and from string representations of basic data types.

os

The os package provides a platform-independent interface to operating system functionality. The design is Unix-like.

sync

The package sync provides basic synchronization primitives such as mutual exclusion locks.

flag

The flag package implements command-line flag parsing.

encoding/json

The encoding/json package implements encoding and decoding of JSON objects as defined in RFC 4627 .

html/template

Data-driven templates for generating textual output such as HTML.

Templates are executed by applying them to a data structure. Annotations in the template refer to elements of the data structure (typically a field of a struct or a key in a map) to control execution and derive values to be displayed. The template walks the structure as it executes and the "cursor" @ represents the value at the current location in the structure.

net/http

The net/http package implements parsing of HTTP requests, replies, and URLs and provides an extensible HTTP server and a basic HTTP client.

unsafe

The unsafe package contains operations that step around the type safety of Go programs. Normally you don't need this package, but it is worth mentioning that unsafe Go programs are possible.

reflect

The reflect package implements run-time reflection, allowing a program to manipulate objects with arbitrary types. The typical use is to take a value with static type interface{} and extract its dynamic type information by calling TypeOf, which returns an object with interface type Type. See , Section .

os/exec

The os/exec package runs external commands.

Exercises

Stack as package

  1. See the Stack exercise. In this exercise we want to create a separate package for that code. Create a proper package for your stack implementation, Push, Pop and the Stack type need to be exported.

  2. Write a simple unit test for this package. You should at least test that a Pop works after a Push.

Answer

  1. There are a few details that should be changed to make a proper package for our stack. First, the exported functions should begin with a capital letter and so should Stack. The package file is named stack-as-package.go and contains:

    package stack
    
    // Stack holds the items.
    type Stack struct {
    	i    int
    	data [10]int
    }
    
    // Push pushes an item on the stack.
    func (s *Stack) Push(k int) {
    	s.data[s.i] = k
    	s.i++
    }
    
    // Pop pops an item from the stack.
    func (s *Stack) Pop() (ret int) {
    	s.i--
    	ret = s.data[s.i]
    	return
    }
    

  2. To make the unit testing work properly you need to do some preparations. We'll come to those in a minute. First the actual unit test. Create a file with the name pushpop_test.go, with the following contents:

    package stack
    
    import "testing"
    
    func TestPushPop(t *testing.T) {
    	c := new(Stack)
    	c.Push(5)
    	if c.Pop() != 5 {
    		t.Log("Pop doesn't give 5")
    		t.Fail()
    	}
    }
    

For go test to work we need to put our package files in a directory under $GOPATH/src:

% mkdir $GOPATH/src/stack
% cp pushpop_test.go $GOPATH/src/stack
% cp stack-as-package.go $GOPATH/src/stack

Yields:

% go test stack
ok stack 0.001s

Calculator

  1. Create a reverse polish calculator. Use your stack package.

Answer

  1. This is one answer:
    package main
    
    import (
    	"bufio"
    	"fmt"
    	"os"
    	"strconv"
    )
    
    var reader *bufio.Reader = bufio.NewReader(os.Stdin)
    var st = new(Stack)
    
    type Stack struct {
    	i    int
    	data [10]int
    }
    
    func (s *Stack) push(k int) {
    	if s.i+1 > 9 {
    		return
    	}
    	s.data[s.i] = k
    	s.i++
    }
    
    func (s *Stack) pop() (ret int) {
    	s.i--
    	if s.i < 0 {
    		s.i = 0
    		return
    	}
    	ret = s.data[s.i]
    	return
    }
    
    func main() {
    	for {
    		s, err := reader.ReadString('\n')
    		var token string
    		if err != nil {
    			return
    		}
    		for _, c := range s {
    			switch {
    			case c >= '0' && c <= '9':
    				token = token + string(c)
    			case c == ' ':
    				r, _ := strconv.Atoi(token)
    				st.push(r)
    				token = ""
    			case c == '+':
    				fmt.Printf("%d\n", st.pop()+st.pop())
    			case c == '*':
    				fmt.Printf("%d\n", st.pop()*st.pop())
    			case c == '-':
    				p := st.pop()
    				q := st.pop()
    				fmt.Printf("%d\n", q-p)
    			case c == 'q':
    				return
    			default:
    				//error
    			}
    		}
    	}
    }
    

Beyond the basics

Go has pointers but not pointer arithmetic. You cannot use a pointer variable to walk through the bytes of a string.

In this chapter we delve deeper in to the language.

Go has pointers. There is however no pointer arithmetic, so they act more like references than pointers that you may know from C. Pointers are useful. Remember that when you call a function in Go, the variables are pass-by-value. So, for efficiency and the possibility to modify a passed value in functions we have pointers.

You declare a pointer by prefixing the type with an '*': var p *int. Now p is a pointer to an integer value. All newly declared variables are assigned their zero value and pointers are no different. A newly declared pointer, or just a pointer that points to nothing, has a nil-value . In other languages this is often called a NULL pointer in Go it is just nil. To make a pointer point to something you can use the address-of operator (&), which we demonstrate here:

var p *int
fmt.Printf("%v", p) 1

var i int	    2
p = &i		    3

fmt.Printf("%v", p) 4

This 1 Prints nil. Declare 2 an integer variable i. Make p point 3 to i, i.e. take the address of i. And this 4 will print something like 0x7ff96b81c000a. De-referencing a pointer is done by prefixing the pointer variable with *.

As said, there is no pointer arithmetic, so if you write: *p++, it is interpreted as (*p)++: first reference and then increment the value.

Allocation

Go also has garbage collection, meaning that you don't have to worry about memory deallocation.13

To allocate memory Go has two primitives, new and make. They do different things and apply to different types, which can be confusing, but the rules are simple. The following sections show how to handle allocation in Go and hopefully clarifies the somewhat artificial distinction between new and make .

Allocation with new

The built-in function new is essentially the same as its namesakes in other languages: new(T) allocates zeroed storage for a new item of type T and returns its address, a value of type *T. Or in other words, it returns a pointer to a newly allocated zero value of type T. This is important to remember.

The documentation for bytes.Buffer states that "the zero value for Buffer is an empty buffer ready to use.". Similarly, sync.Mutex does not have an explicit constructor or Init method. Instead, the zero value for a sync.Mutex is defined to be an unlocked mutex.

Allocation with make

The built-in function make(T, args) serves a purpose different from new(T). It creates slices, maps, and channels only, and it returns an initialized (not zero!) value of type T, and not a pointer: *T. The reason for the distinction is that these three types are, under the covers, references to data structures that must be initialized before use. A slice, for example, is a three-item descriptor containing a pointer to the data (inside an array), the length, and the capacity; until those items are initialized, the slice is nil. For slices, maps, and channels, make initializes the internal data structure and prepares the value for use.

For instance, make([]int, 10, 100) allocates an array of 100 ints and then creates a slice structure with length 10 and a capacity of 100 pointing at the first 10 elements of the array. In contrast, new([]int) returns a pointer to a newly allocated, zeroed slice structure, that is, a pointer to a nil slice value. These examples illustrate the difference between new and make.

var p *[]int = new([]int)       1
var v  []int = make([]int, 100) 2

var p *[]int = new([]int)       3
*p = make([]int, 100, 100)

v := make([]int, 100)           4

Allocates 1 slice structure; rarely useful. v 2 refers to a new array of 100 ints. At 3 we make it unnecessarily complex, 4 is more idiomatic.

Remember that make applies only to maps, slices and channels and does not return a pointer. To obtain an explicit pointer allocate with new.

Constructors and composite literals

Sometimes the zero value isn't good enough and an initializing constructor is necessary, as in this example taken from the package os.

func NewFile(fd int, name string) *File {
    if fd < 0 {
        return nil
    }
    f := new(File)
    f.fd = fd
    f.name = name
    f.dirinfo = nil
    f.nepipe = 0
    return f
}

There's a lot of boiler plate in there. We can simplify it using a composite literal , which is an expression that creates a new instance each time it is evaluated.

func NewFile(fd int, name string) *File {
    if fd < 0 {
        return nil
    }
    f := File{fd, name, nil, 0}
    return &f	1
}

It is OK to return the address of a local variable 1 the storage associated with the variable survives after the function returns.

In fact, taking the address of a composite literal allocates a fresh instance each time it is evaluated, so we can combine these last two lines.14

return &File{fd, name, nil, 0}

The items (called fields) of a composite literal are laid out in order and must all be present. However, by labeling the elements explicitly as field:value pairs, the initializers can appear in any order, with the missing ones left as their respective zero values. Thus we could say

return &File{fd: fd, name: name}

As a limiting case, if a composite literal contains no fields at all, it creates a zero value for the type. The expressions new(File) and &File{} are equivalent. In fact the use of new is discouraged.

Composite literals can also be created for arrays, slices, and maps, with the field labels being indices or map keys as appropriate. In these examples, the initializations work regardless of the values of Enone, and Einval, as long as they are distinct:

ar := [...]string{Enone: "no error", Einval: "invalid argument"}
sl := []string{Enone: "no error", Einval: "invalid argument"}
ma := map[int]string {Enone: "no error", Einval: "invalid argument"}

Defining your own types

Of course Go allows you to define new types, it does this with the type keyword: type foo int

This creates a new type foo which acts like an int. Creating more sophisticated types is done with the struct keyword. An example would be when we want record somebody's name (string) and age (int) in a single structure and make it a new type:

package main

import "fmt"

type NameAge struct {
	name string // Both non exported fields.
	age  int
}

func main() {
	a := new(NameAge)
	a.name = "Pete"
	a.age = 42
	fmt.Printf("%v\n", a)
}

Apropos, the output of fmt.Printf("%v\n", a) is &{Pete 42}

That is nice! Go knows how to print your structure. If you only want to print one, or a few, fields of the structure you'll need to use .<field name>. For example to only print the name:

fmt.Printf("%s", a.name)

More on structure fields

As said each item in a structure is called a field. A struct with no fields: struct {}. Or one with four fields:

struct {
    x, y int
    A *[]int
    F func()
}

If you omit the name for a field, you create an anonymous field , for instance:

struct {
    T1        // Field name is T1.
    *T2       // Field name is T2.
    P.T3      // Field name is T3.
    x, y int  // Field names are x and y.
}

Note that field names that start with a capital letter are exported, i.e. can be set or read from other packages. Field names that start with a lowercase are private to the current package. The same goes for functions defined in packages, see for the details.

Methods

If you create functions that work on your newly defined type, you can take two routes:

  1. Create a function that takes the type as an argument.
func doSomething(n1 *NameAge, n2 int) { /* */ }
  1. Create a function that works on the type (see receiver in ):
func (n1 *NameAge) doSomething(n2 int) { /* */ }

This is a method call, which can be used as:

var n *NameAge
n.doSomething(2)

Whether to use a function or method is entirely up to the programmer, but if you want to satisfy an interface (see the next chapter) you must use methods. If no such requirement exists it is a matter of taste whether to use functions or methods.

But keep the following in mind, this is quoted from :

If x is addressable and &x's method set contains m, x.m() is shorthand for (&x).m().

In the above case this means that the following is not an error:

var n NameAge	    // Not a pointer
n.doSomething(2)

Here Go will search the method list for n of type NameAge, come up empty and will then also search the method list for the type *NameAge and will translate this call to (&n).doSomething(2).

There is a subtle but major difference between the following type declarations. Also see the Section "Type Declarations" . Suppose we have:

// A Mutex is a data type with two methods, Lock and Unlock.
type Mutex struct         { /* Mutex fields */ }
func (m *Mutex) Lock()    { /* Lock impl. */ }
func (m *Mutex) Unlock()  { /* Unlock impl. */ }

We now create two types in two different manners:

NewMutex is equal to Mutex, but it does not have any of the methods of Mutex. In other words its method set is empty. But PrintableMutex has inherited the method set from Mutex. The Go term for this is embedding . In the words of :

The method set of *PrintableMutex contains the methods Lock and Unlock bound to its anonymous field Mutex.

Conversions

Sometimes you want to convert a type to another type. This is possible in Go, but there are some rules. For starters, converting from one value to another is done by operators (that look like functions: byte()) and not all conversions are allowed.

Valid conversions, float64 works the same as float32.
From b []byte i []int r []rune s string f float32 i int
To
[]byte · []byte(s)
[]int · []int(s)
[]rune []rune(s)
string string(b) string(i) string(r) ·
float32 · float32(i)
int int(f) ·

For numeric values the following conversions are defined:

User defined types and conversions

How can you convert between the types you have defined yourself? We create two types here Foo and Bar, where Bar is an alias for Foo:

type foo struct { int }  // Anonymous struct field.
type bar foo             // bar is an alias for foo.

Then we:

var b bar = bar{1} // Declare `b` to be a `bar`.
var f foo = b	   // Assign `b` to `f`.

Which fails on the last line with: cannot use b (type bar) as type foo in assignment

This can be fixed with a conversion: var f foo = foo(b)

Note that converting structures that are not identical in their fields is more difficult. Also note that converting b to a plain int also fails; an integer is not the same as a structure containing an integer.

Exercises

Map function with interfaces

  1. Use the answer from the earlier map exercise but now make it generic using interfaces. Make it at least work for ints and strings.

Answer

  1. package main
    
    import "fmt"
    
    // Define the empty interface as a type.
    type e interface{}
    
    func mult2(f e) e {
    	switch f.(type) {
    	case int:
    		return f.(int) * 2
    	case string:
    		return f.(string) + f.(string) + f.(string) + f.(string)
    	}
    	return f
    }
    
    func Map(n []e, f func(e) e) []e {
    	m := make([]e, len(n))
    	for k, v := range n {
    		m[k] = f(v)
    	}
    	return m
    }
    
    func main() {
    	m := []e{1, 2, 3, 4}
    	s := []e{"a", "b", "c", "d"}
    	mf := Map(m, mult2)
    	sf := Map(s, mult2)
    	fmt.Printf("%v\n", mf)
    	fmt.Printf("%v\n", sf)
    }
    

Pointers

  1. Suppose we have defined the following structure:

    type Person struct {
        name string
        age  int
    }
    

    What is the difference between the following two lines?

    var p1 Person
    p2 := new(Person)
    
  2. What is the difference between the following two allocations?

    func Set(t *T) {
        x = t
    }
    

    and

    func Set(t T) {
        x= &t
    }
    

Answer

  1. The expression, var p1 Person allocates a Person-value to p1. The type of p1 is Person. The second line: p2 := new(Person) allocates memory and assigns a pointer to p2. The type of p2 is *Person.

  2. In the first function, x points to the same thing that t does, which is the same thing that the actual argument points to. So in the second function, we have an "extra" variable containing a copy of the interesting value. In the second function, x points to a new (heap-allocated) variable t which contains a copy of whatever the actual argument value is.

Linked List

  1. Make use of the package container/list to create a (doubly) linked list. Push the values 1, 2 and 4 to the list and then print it.

  2. Create your own linked list implementation. And perform the same actions as above.

Answer

  1. The following is the implementation of a program using doubly linked lists from container/list.

    package main
    
    import (
    	"container/list"
    	"fmt"
    )
    
    func main() {
    	l := list.New()
    	l.PushBack(1)
    	l.PushBack(2)
    	l.PushBack(4)
    
    	for e := l.Front(); e != nil; e = e.Next() {
    		fmt.Printf("%v\n", e.Value)
    	}
    }
    

  2. The following is a program implementing a simple doubly linked list supporting int values.

    package main
    
    import (
    	"errors" 1
    	"fmt"
    )
    
    type Value int 2
    
    type Node struct { 3
    	Value
    	prev, next *Node
    }
    
    type List struct {
    	head, tail *Node
    }
    
    func (l *List) Front() *Node { 4
    	return l.head
    }
    
    func (n *Node) Next() *Node {
    	return n.next
    }
    
    func (l *List) Push(v Value) *List {
    	n := &Node{Value: v} 5
    
    	if l.head == nil { 6
    		l.head = n
    	} else {
    		l.tail.next = n 7
    		n.prev = l.tail 8
    	}
    	l.tail = n 9
    
    	return l
    }
    
    var errEmpty = errors.New("List is empty")
    
    func (l *List) Pop() (v Value, err error) {
    	if l.tail == nil { 10
    		err = errEmpty
    	} else {
    		v = l.tail.Value     11
    		l.tail = l.tail.prev 12
    		if l.tail == nil {
    			l.head = nil 13
    		}
    	}
    
    	return v, err
    }
    
    func main() {
    	l := new(List)
    
    	l.Push(1)
    	l.Push(2)
    	l.Push(4)
    
    	for n := l.Front(); n != nil; n = n.Next() {
    		fmt.Printf("%v\n", n.Value)
    	}
    
    	fmt.Println()
    
    	for v, err := l.Pop(); err == nil; v, err = l.Pop() {
    		fmt.Printf("%v\n", v)
    	}
    }
    

Import 1 the packages we will need. At 2 we declare a type for the value our list will contain, this is not strictly neccesary. And at 3 we declare a type for the each node in our list. At 4 we define the Front method for our list. When pushing, create a new Node 5 with the provided value. If the list is empty 6, put the new node at the head. Otherwise 7 put it at the tail and make sure 8 the new node points back to the previously existing one. At 9 we re-adjust tail to the newly inserted node.

In the Pop 10 method, we return an error if the list is empty. If it is not empty 11 we save the last value. And then 12 discard the last node from the list. Finally at 13 we make sure the list is consistent if it becomes empty.

Cat

  1. Write a program which mimics the Unix program cat.

  2. Make it support the -n flag, where each line is numbered.

  3. The solution to the above question given in contains a bug. Can you spot and fix it?

Answer

  1. The following is implemention of cat which also supports a -n flag to number each line.

package main

import (
	"bufio"
	"flag"
	"fmt"
	"io" 1
	"os"
)

var numberFlag = flag.Bool("n", false, "number each line")  2

func cat(r *bufio.Reader) { 3
	i := 1
	for {
		buf, e := r.ReadBytes('\n') 4
		if e == io.EOF {            5
			break
		}
		if *numberFlag { 6
			fmt.Fprintf(os.Stdout, "%5d  %s", i, buf)
			i++
		} else { 7
			fmt.Fprintf(os.Stdout, "%s", buf)
		}
	}
	return
}

func main() {
	flag.Parse()
	if flag.NArg() == 0 {
		cat(bufio.NewReader(os.Stdin))
	}
	for i := 0; i < flag.NArg(); i++ {
		f, e := os.Open(flag.Arg(i))
		if e != nil {
			fmt.Fprintf(os.Stderr, "%s: error reading from %s: %s\n",
				os.Args[0], flag.Arg(i), e.Error())
			continue
		}
		cat(bufio.NewReader(f))
	}
}

At 1 we include all the packages we need. Here 2 we define a new flag "n", which defaults to off. Note that we get the help (-h) for free. Start the function 3 that actually reads the file's contents and displays it; Read one line at the time at 4. And stop 5 if we hit the end. If we should number each line, print the line number and then the line itself 6. Otherwise 7 we could just print the line.

  1. The bug show itself when the last line of the input does not contain a newline. Or worse, when the input contains one line without a closing newline nothing is shown at all. A better solution is the following program.
    package main
    
    import (
    	"bufio"
    	"flag"
    	"fmt"
    	"io"
    	"os"
    )
    
    var numberFlag = flag.Bool("n", false, "number each line")
    
    func cat(r *bufio.Reader) {
    	i := 1
    	for {
    		buf, e := r.ReadBytes('\n')
    		if e == io.EOF && string(buf) == "" {
    			break
    		}
    		if *numberFlag {
    			fmt.Fprintf(os.Stdout, "%5d  %s", i, buf)
    			i++
    		} else {
    			fmt.Fprintf(os.Stdout, "%s", buf)
    		}
    	}
    	return
    }
    
    func main() {
    	flag.Parse()
    	if flag.NArg() == 0 {
    		cat(bufio.NewReader(os.Stdin))
    	}
    	for i := 0; i < flag.NArg(); i++ {
    		f, e := os.Open(flag.Arg(i))
    		if e != nil {
    			fmt.Fprintf(os.Stderr, "%s: error reading from %s: %s\n",
    				os.Args[0], flag.Arg(i), e.Error())
    			continue
    		}
    		cat(bufio.NewReader(f))
    	}
    }
    

Method calls

  1. Suppose we have the following program. Note the package container/vector was once part of Go, but was removed when the append built-in was introduced. However, for this question this isn't important. The package implemented a stack-like structure, with push and pop methods.

    package main
    
    import "container/vector"
    
    func main() {
        k1 := vector.IntVector{}
        k2 := &vector.IntVector{}
        k3 := new(vector.IntVector)
        k1.Push(2)
        k2.Push(3)
        k3.Push(4)
    }
    

    What are the types of k1, k2 and k3?

  2. Now, this program compiles and runs OK. All the Push operations work even though the variables are of a different type. The documentation for Push says:

    func (p *IntVector) Push(x int) Push appends x to the end of the vector.

    So the receiver has to be of type *IntVector, why does the code above (the Push statements) work correctly then?

Answer

  1. The type of k1 is vector.IntVector. Why? We use a composite literal (the {}), so we get a value of that type back. The variable k2 is of *vector.IntVector, because we take the address (&) of the composite literal. And finally k3 has also the type *vector.IntVector, because new returns a pointer to the type.

  2. The answer is given in in the section "Calls", where among other things it says:

A method call x.m() is valid if the method set of (the type of) x contains m and the argument list can be assigned to the parameter list of m. If x is addressable and &x's method set contains m, x.m() is shorthand for (&x).m().

In other words because k1 is addressable and *vector.IntVector does have the Push method, the call k1.Push(2) is translated by Go into (&k1).Push(2) which makes the type system happy again (and you too -- now you know this).15

Interfaces

I have this phobia about having my body penetrated surgically. You know what I mean?

In Go, the word interface is overloaded to mean several different things. Every type has an interface, which is the set of methods defined for that type. This bit of code defines a struct type S with one field, and defines two methods for S. 16

type S struct { i int }
func (p *S) Get() int  { return p.i }
func (p *S) Put(v int) { p.i = v }
Defining a struct and methods on it.

You can also define an interface type, which is simply a set of methods. This defines an interface I with two methods:

type I interface {
    Get() int
    Put(int)
}

S is a valid implementation for interface I, because it defines the two methods which I requires. Note that this is true even though there is no explicit declaration that S implements I.

A Go program can use this fact via yet another meaning of interface, which is an interface value:

func f(p I) { 1
    fmt.Println(p.Get()) 2
    p.Put(1) 3
}

At 1 we declare a function that takes an interface type as the argument. Because p implements I, it must have the Get() method, which we call at 2. And the same holds true for the Put() method at 3. Because S implements I, we can call the function f passing in a pointer to a value of type S: var s S; f(&s)

The reason we need to take the address of s, rather than a value of type S, is because we defined the methods on s to operate on pointers, see the definition in the code above. This is not a requirement -- we could have defined the methods to take values -- but then the Put method would not work as expected.

The fact that you do not need to declare whether or not a type implements an interface means that Go implements a form of duck typing . This is not pure duck typing, because when possible the Go compiler will statically check whether the type implements the interface. However, Go does have a purely dynamic aspect, in that you can convert from one interface type to another. In the general case, that conversion is checked at run time. If the conversion is invalid -- if the type of the value stored in the existing interface value does not satisfy the interface to which it is being converted -- the program will fail with a run time error.

Interfaces in Go are similar to ideas in several other programming languages: pure abstract virtual base classes in C++, typeclasses in Haskell or duck typing in Python. However there is no other language which combines interface values, static type checking, dynamic run time conversion, and no requirement for explicitly declaring that a type satisfies an interface. The result in Go is powerful, flexible, efficient, and easy to write.

Which is what?

Let's define another type R that also implements the interface I:

type R struct { i int }
func (p *R) Get() int  { return p.i }
func (p *R) Put(v int) { p.i = v }

The function f can now accept variables of type R and S.

Suppose you need to know the actual type in the function f. In Go you can figure that out by using a type switch.

func f(p I) {
    switch t := p.(type) { 1
        case *S: 2
        case *R: 3
        default: 4
    }
}

At 1 we use the type switch, note that the .(type) syntax is only valid within a switch statement. We store the value in the variable t. The subsequent cases 2 3 each check for a different actual type. And we can even have a default 4 clause. It is worth pointing out that both case R and case s aren't possible, because p needs to be a pointer in order to satisfy i.

A type switch isn't the only way to discover the type at run-time.

if t, ok := something.(I); ok { 1
     ...
}

You can also use a "comma, ok" form 1 to see if an interface type implements a specific interface. If ok is true, t will hold the type of something. When you are sure a variable implements an interface you can use: t := something.(I) .

Empty interface

Since every type satisfies the empty interface: interface{} we can create a generic function which has an empty interface as its argument:

func g(something interface{}) int {
    return something.(I).Get()
}

The return something.(I).Get() is the tricky bit in this function. The value something has type interface{}, meaning no guarantee of any methods at all: it could contain any type. The .(I) is a type assertion which converts something to an interface of type I. If we have that type we can invoke the Get() function. So if we create a new variable of the type *S, we can just call g(), because *S also implements the empty interface.

s = new(S)
fmt.Println(g(s));

The call to g will work fine and will print 0. If we however invoke g() with a value that does not implement I we have a problem:

var i int
fmt.Println(g(i))

This compiles, but when we run this we get slammed with: "panic: interface conversion: int is not main.I: missing method Get".

Which is completely true, the built-in type int does not have a Get() method.

Methods

Methods are functions that have a receiver (see ). You can define methods on any type (except on non-local types, this includes built-in types: the type int can not have methods). You can however make a new integer type with its own methods. For example:

type Foo int

func (self Foo) Emit() {
    fmt.Printf("%v", self)
}

type Emitter interface {
    Emit()
}

Doing this on non-local (types defined in other packages) types yields an error "cannot define new methods on non-local type int".

Methods on interface types

An interface defines a set of methods. A method contains the actual code. In other words, an interface is the definition and the methods are the implementation. So a receiver can not be an interface type, doing so results in a "invalid receiver type ..." compiler error. The authoritative word from the language spec :

The receiver type must be of the form T or *T where T is a type name. T is called the receiver base type or just base type. The base type must not be a pointer or interface type and must be declared in the same package as the method.

Interface names

By convention, one-method interfaces are named by the method name plus the -er suffix: Reader, Writer, Formatter etc.

There are a number of such names and it's productive to honor them and the function names they capture. Read, Write, Close, Flush, String and so on have canonical signatures and meanings. To avoid confusion, don't give your method one of those names unless it has the same signature and meaning. Conversely, if your type implements a method with the same meaning as a method on a well-known type, give it the same name and signature; call your string-converter method String not ToString. 17

A sorting example

Recall the Bubblesort exercise, where we sorted an array of integers:

func bubblesort(n []int) {
    for i := 0; i < len(n)-1; i++ {
        for j := i + 1; j < len(n); j++ {
            if n[j] < n[i] {
                n[i], n[j] = n[j], n[i]
            }
        }
    }
}

A version that sorts strings is identical except for the signature of the function: func bubblesortString(n []string) { /* ... */ } . Using this approach would lead to two functions, one for each type. By using interfaces we can make this more generic. Let's create a new function that will sort both strings and integers, something along the lines of this non-working example:

func sort(i []interface{}) {  1
    switch i.(type) {         2
    case string:              3
         ...
    case int:
         ...
    }
    return /* ... */          4
}

Our function will receive a slice of empty interfaces at 1. We then 2 use a type switch to find out what the actual type of the input is. And then 3 then sort accordingly. And, when done, return 4 the sorted slice.

But when we call this function with sort([]int{1, 4, 5}), it fails with: "cannot use i (type []int) as type []interface in function argument"

This is because Go can not easily convert to a slice of interfaces. Just converting to an interface is easy, but to a slice is much more costly. The full mailing list discussion on this subject can be found at . To keep a long story short: Go does not (implicitly) convert slices for you.

So what is the Go way of creating such a "generic" function? Instead of doing the type inference ourselves with a type switch, we let Go do it implicitly: The following steps are required:

Listing interfaces in interfaces

Take a look at the following example of an interface definition, this one is from the package container/heap:

type Interface interface {
    sort.Interface
    Push(x interface{})
    Pop() interface{}
}

Here another interface is listed inside the definition of heap.Interface, this may look odd, but is perfectly valid, remember that on the surface an interface is nothing more than a listing of methods. sort.Interface is also such a listing, so it is perfectly legal to include it in the interface.

Introspection and reflection

In the following example we want to look at the "tag" (here named "namestr") defined in the type definition of Person. To do this we need the reflect package (there is no other way in Go). Keep in mind that looking at a tag means going back to the type definition. So we use the reflect package to figure out the type of the variable and then access the tag.

type Person struct {
    name string "namestr"
    age  int
}

func ShowTag(i interface{}) { 1
    switch t := reflect.TypeOf(i); t.Kind() {
    case reflect.Ptr: 2
        tag := t.Elem().Field(0).Tag
                 3     4       5
Introspection using reflection.

We are calling ShowTag at 1 with a *Person, so at 2 we're expecting a reflect.Ptr. We are dealing with a Type 3 and according to the documentation 18:

Elem returns a type's element type. It panics if the type's Kind is not Array, Chan, Map, Ptr, or Slice.

So on t we use Elem() to get the value the pointer points to. We have now dereferenced the pointer and are "inside" our structure. We then 4 use Field(0) to access the zeroth field.

The struct StructField has a Tag member which returns the tag-name as a string. So on the \(0^{th}\) field we can unleash .Tag 5 to access this name: Field(0).Tag. This gives us namestr.

To make the difference between types and values more clear, take a look at the following code:

func show(i interface{}) {
    switch t := i.(type) {
    case *Person:
        t := reflect.TypeOf(i)  1
        v := reflect.ValueOf(i) 2
        tag := t.Elem().Field(0).Tag 3
        name := v.Elem().Field(0).String() 4
    }
}
Reflection and the type and value.

At 1 we create t the type data of i, and v gets the actual values at 2. Here at 3 we want to get to the "tag". So we need Elem() to redirect the pointer, access the first field and get the tag. Note that we operate on t a reflect.Type. Now 4 we want to get access to the value of one of the members and we employ Elem() on v to do the redirection. we have "arrived" at the structure. Then we go to the first field Field(0) and invoke the String() method on it.

Peeling away the layers using reflection.
Peeling away the layers using reflection. Going from a *Person via Elem using the methods described in go doc reflect to get the actual string contained within.

Setting a value works similarly as getting a value, but only works on exported members. Again some code:

type Person struct {
    name string
    age  int
}

func Set(i interface{}) {
    switch i.(type) {
    case *Person:
        r := reflect.ValueOf(i)
        r.Elem(0).Field(0).SetString("Albert Einstein")
    }
}
Reflect with private member.
type Person struct {
    Name string
    age  int
}

func Set(i interface{}) {
    switch i.(type) {
    case *Person:
        r := reflect.ValueOf(i)
        r.Elem().Field(0).SetString("Albert Einstein")
    }
}
Reflect with public member.

The first program compiles and runs, but when you run it, you are greeted with a stack trace and a run time error: "panic: reflect.Value.SetString using value obtained using unexported field".

The second program works OK and sets the member Name to "Albert Einstein". Of course this only works when you call Set() with a pointer argument.

Exercises

Interfaces and max()

In the maximum exercise we created a max function that works on a slice of integers. The question now is to create a program that shows the maximum number and that works for both integers and floats. Try to make your program as generic as possible, although that is quite difficult in this case.

Answer

The following program calculates a maximum. It is as generic as you can get with Go.

package main

import "fmt"

func Less(l, r interface{}) bool { 1
    switch l.(type) {
    case int:
        if _, ok := r.(int); ok {
            return l.(int) < r.(int) 2
        }
    case float32:
        if _, ok := r.(float32); ok {
            return l.(float32) < r.(float32) 3
        }
    }
    return false
}

func main() {
    var a, b, c int = 5, 15, 0
    var x, y, z float32 = 5.4, 29.3, 0.0

    if c = a; Less(a, b) { 4
        c = b
    }
    if z = x; Less(x, y) { 5
        z = y
    }
    fmt.Println(c, z)
}

We could have chosen to make the return type of this 1 function an interface{}, but that would mean that a caller would always have to do a type assertion to extract the actual type from the interface. At 2 we compare the parameters. All parameters are confirmed to be integers, so this is legit. And at 3 we do the some for floats. At 4 5 we get the maximum value for a, b and x and y.

Pointers and reflection

One of the last paragraphs in section "\titleref{sec:introspection and reflection}" on page \pageref{sec:introspection and reflection}, has the following words:

The code on the right works OK and sets the member Name to "Albert Einstein". Of course this only works when you call Set() with a pointer argument.

Why is this the case?

Answer

When called with a non-pointer argument the variable is a copy (call-by-value). So you are doing the reflection voodoo on a copy. And thus you are not changing the original value, but only this copy.

Concurrency

In this chapter we will show off Go's ability for concurrent programming using channels and goroutines. Goroutines are the central entity in Go's ability for concurrency.

But what is a goroutine, from :

They're called goroutines because the existing terms -- threads, coroutines, processes, and so on -- convey inaccurate connotations. A goroutine has a simple model: it is a function executing in parallel with other goroutines in the same address space. It is lightweight, costing little more than the allocation of stack space. And the stacks start small, so they are cheap, and grow by allocating (and freeing) heap storage as required.

A goroutine is a normal function, except that you start it with the keyword go.

ready("Tea", 2)	    // Normal function call.
go ready("Tea", 2)  // ... as goroutine.

func ready(w string, sec int) {
	time.Sleep(time.Duration(sec) * time.Second)
	fmt.Println(w, "is ready!")
}

func main() {
	go ready("Tea", 2) 1
	go ready("Coffee", 1) 2
	fmt.Println("I'm waiting")
	time.Sleep(5 * time.Second) 3
}
Go routines in action.

The following idea for a program was taken from . We run a function as two goroutines, the goroutines wait for an amount of time and then print something to the screen. At 1 2 we start the goroutines. The main function waits long enough at 3, so that both goroutines will have printed their text. Right now we wait for 5 seconds, but in fact we have no idea how long we should wait until all goroutines have exited. This outputs:

I'm waiting         // Right away
Coffee is ready!    // After 1 second
Tea is ready!       // After 2 seconds

If we did not wait for the goroutines (i.e. remove the last line at <2>) the program would be terminated immediately and any running goroutines would die with it.

To fix this we need some kind of mechanism which allows us to communicate with the goroutines. This mechanism is available to us in the form of channels . A channel can be compared to a two-way pipe in Unix shells: you can send to and receive values from it. Those values can only be of a specific type: the type of the channel. If we define a channel, we must also define the type of the values we can send on the channel. Note that we must use make to create a channel:

ci := make(chan int)
cs := make(chan string)
cf := make(chan interface{})

Makes ci a channel on which we can send and receive integers, makes cs a channel for strings and cf a channel for types that satisfy the empty interface. Sending on a channel and receiving from it, is done with the same operator: <-.

Depending on the operands it figures out what to do:

ci <- 1   // *Send* the integer 1 to the channel ci.
<-ci      // *Receive* an integer from the channel ci.
i := <-ci // *Receive* from the channel ci and store it in i.

Let's put this to use.

var c chan int 1

func ready(w string, sec int) {
    time.Sleep(time.Duration(sec) * time.Second)
    fmt.Println(w, "is ready!")
    c <- 1	2
}

func main() {
    c = make(chan int) 3
    go ready("Tea", 2) 4
    go ready("Coffee", 1) 5
    fmt.Println("I'm waiting, but not too long")
    <-c 6
    <-c 7
}

At 1 we declare c to be a variable that is a channel of ints. That is: this channel can move integers. Note that this variable is global so that the goroutines have access to it. At 2 in the ready function we send the integer 1 on the channel. In our main function we initialize c at 3 and start our goroutines 4 5. At 6 7 we Wait until we receive a value from the channel, the value we receive is discarded. We have started two goroutines, so we expect two values to receive.

There is still some remaining ugliness; we have to read twice from the channel 6 7). This is OK in this case, but what if we don't know how many goroutines we started? This is where another Go built-in comes in: select . With select you can (among other things) listen for incoming data on a channel.

Using select in our program does not really make it shorter, because we run too few go-routines. We remove last lines and replace them with the following:

L: for {
    select {
    case <-c:
        i++
        if i > 1 {
            break L
        }
    }
}

We will now wait as long as it takes. Only when we have received more than one reply on the channel c will we exit the loop L.

Make it run in parallel

While our goroutines were running concurrently, they were not running in parallel. When you do not tell Go anything there can only be one goroutine running at a time. With runtime.GOMAXPROCS(n) you can set the number of goroutines that can run in parallel. From the documentation:

GOMAXPROCS sets the maximum number of CPUs that can be executing simultaneously and returns the previous setting. If n < 1, it does not change the current setting. This call will go away when the scheduler improves.

If you do not want to change any source code you can also set an environment variable GOMAXPROCS to the desired value.

Note that the above discussion relates to older versions of Go. From version 1.5 and above, GOMAXPROCS defaults to the number of CPU cores.

More on channels

When you create a channel in Go with ch := make(chan bool), an unbuffered channel for bools is created. What does this mean for your program? For one, if you read (value := <-ch) it will block until there is data to receive. Secondly anything sending (ch <- true) will block until there is somebody to read it. Unbuffered channels make a perfect tool for synchronizing multiple goroutines.

But Go allows you to specify the buffer size of a channel, which is quite simply how many elements a channel can hold. ch := make(chan bool, 4), creates a buffered channel of bools that can hold 4 elements. The first 4 elements in this channel are written without any blocking. When you write the 5th element, your code will block, until another goroutine reads some elements from the channel to make room.

In conclusion, the following is true in Go:

\[ \textsf{ch := make(chan type, value)} \left\{ \begin{array}{ll} value == 0 & \rightarrow \textsf{unbuffered} \\ value > 0 & \rightarrow \textsf{buffer }{} value{} \textsf{ elements} \end{array} \right. \]

When a channel is closed the reading side needs to know this. The following code will check if a channel is closed.

x, ok = <-ch

Where ok is set to true the channel is not closed and we've read something. Otherwise ok is set to false. In that case the channel was closed and the value received is a zero value of the channel's type.

Exercises

Channels

  1. Modify the program you created in exercise to use channels, in other words, the function called in the body should now be a goroutine and communication should happen via channels. You should not worry yourself on how the goroutine terminates.

  2. There are a few annoying issues left if you resolve question 1 above. One of the problems is that the goroutine isn't neatly cleaned up when main.main() exits. And worse, due to a race condition between the exit of main.main() and main.shower() not all numbers are printed. It should print up until 9, but sometimes it prints only to 8. Adding a second quit-channel you can remedy both issues. Do this.

Answer

  1. A possible program is:

package main

import "fmt"

func main() {
	ch := make(chan int)
	go shower(ch)
	for i := 0; i < 10; i++ {
		ch <- i
	}
}

func shower(c chan int) {
	for {
		j := <-c
		fmt.Printf("%d\n", j)
	}
}

We start in the usual way, then at line 6 we create a new channel of ints. In the next line we fire off the function shower with the ch variable as it argument, so that we may communicate with it. Next we start our for-loop (lines 8-10) and in the loop we send (with <-) our number to the function (now a goroutine) shower.

In the function shower we wait (as this blocks) until we receive a number (line 15). Any received number is printed (line 16) and then continue the endless loop started on line 14.

  1. An answer is

package main

import "fmt"

func main() {
	ch := make(chan int)
	quit := make(chan bool)
	go shower(ch, quit)
	for i := 0; i < 10; i++ {
		ch <- i
	}
	quit <- false	// or true, does not matter
}

func shower(c chan int, quit chan bool) {
	for {
		select {
		case j := <-c:
			fmt.Printf("%d\n", j)
		case <-quit:
			break
		}
	}
}

On line 20 we read from the quit channel and we discard the value we read. We could have used q := <-quit, but then we would have used the variable only once --- which is illegal in Go. Another trick you might have pulled out of your hat may be: _ = <-quit. This is valid in Go, but idomatic Go is the one given on line 20.

Fibonacci II

This is the same exercise as an earlier one in exercise. For completeness the complete question:

The Fibonacci sequence starts as follows: \(1, 1, 2, 3, 5, 8, 13, \ldots\) Or in mathematical terms: \( x_1 = 1; x_2 = 1; x_n = x_{n-1} + x_{n-2}\quad\forall n > 2 \).

Write a function that takes an int value and gives that many terms of the Fibonacci sequence.

But now the twist: You must use channels.

Answer

The following program calculates the Fibonacci numbers using channels.

package main
import "fmt"

func dup3(in <-chan int) (<-chan int, <-chan int, <-chan int) {
	a, b, c := make(chan int, 2), make(chan int, 2), make(chan int, 2)
	go func() {
		for {
			x := <-in
			a <- x
			b <- x
			c <- x
		}
	}()
	return a, b, c
}

func fib() <-chan int {
	x := make(chan int, 2)
	a, b, out := dup3(x)
	go func() {
		x <- 0
		x <- 1
		<-a
		for {
			x <- <-a+<-b
		}
	}()
	return out
}

func main() {
	x := fib()
	for i := 0; i < 10; i++ {
		fmt.Println(<-x)
	}
}

// See sdh33b.blogspot.com/2009/12/fibonacci-in-go.html

Communication

Good communication is as stimulating as black coffee, and just as hard to sleep after.

In this chapter we are going to look at the building blocks in Go for communicating with the outside world. We will look at files, directories, networking and executing other programs. Central to Go's I/O are the interfaces io.Reader and io.Writer. The io.Reader interface specifies one method Read(p []byte) (n int, err err).

Reading from (and writing to) files is easy in Go. This program only uses the os package to read data from the file /etc/passwd.

package main

import (
	"log"
	"os"
)

func main() {
	buf := make([]byte, 1024)
	f, e := os.Open("/etc/passwd") 1
	if e != nil {
		log.Fatalf(e)
	}
	defer f.Close()                2
	for {
		n, e := f.Read(buf) 3
		if e != nil {
			log.Fatalf(e) 4
		}
		if n == 0 { 5
			break
		}
		os.Stdout.Write(buf[:n]) 6
	}
}

We open the file at 1 with os.Open that returns a *os.File *os.File implements io.Reader and io.Writer interface. After the Open we directly put the f.Close() which we defer until the function return. At 3 we call Read on f and read up to 1024 bytes at the time. If anything fails we bail out at 4. If the number of bytes read is 0 we've read the end of the file 5. And at 6 we output the buffer to standard output.

If you want to use buffered I/O there is the bufio package:

package main

import (
	"bufio"
	"log"
	"os"
)

func main() {
	buf := make([]byte, 1024)
	f, e := os.Open("/etc/passwd") 1
	if e != nil {
		log.Fatalf(e)
	}
	defer f.Close()
	r := bufio.NewReader(f) 2
	w := bufio.NewWriter(os.Stdout)
	defer w.Flush() 3
	for {
		n, e := r.Read(buf) 4
		if e != nil {
			log.Fatalf(e)
		}
		if n == 0 {
			break
		}
		w.Write(buf[0:n]) 5
	}
}

Again, we open 1 the file. Then at 2 we Turn f into a buffered Reader. NewReader expects an io.Reader, so you this will work. Then at 4 we read and at 5 we write. We also call Flush() at 3 to flush all output. This entire program could be optimized further by using io.Copy.

io.Reader

As mentioned above the io.Reader is an important interface in the language Go. A lot (if not all) functions that need to read from something take an io.Reader as input. To fulfill the interface a type needs to implement that one method. The writing side io.Writer, has the Write method.

If you think of a new type in your program or package and you make it fulfill the io.Reader or io.Writer interface, the whole standard Go library can be used on that type!

Some examples

The previous program reads a file in its entirety, but a more common scenario is that you want to read a file on a line-by-line basis. The following snippet shows a way to do just that (we're discarding the error returned from os.Open here to keep the examples smaller -- don't ever do this in real life code).

f, _ := os.Open("/etc/passwd"); defer f.Close()
r := bufio.NewReader(f) 1
s, ok := r.ReadString('\n') 2

At 1 make f a bufio to have access to the ReadString method. Then at 2 we read a line from the input, s now holds a string which we can manipulate with, for instance, the strings package.

A more robust method (but slightly more complicated) is ReadLine, see the documentation of the bufio package.

A common scenario in shell scripting is that you want to check if a directory exists and if not, create one.

if [ ! -e name ]; then          if f, e := os.Stat("name"); e != nil {
    mkdir name                      os.Mkdir("name", 0755)
else                            } else {
    # error                         // error
fi                              }

The similarity between these two examples (and with other scripting languages) have prompted comments that Go has a "script"-like feel to it, i.e. programming in Go can be compared to programming in an interpreted language (Python, Ruby, Perl or PHP).

Command line arguments

Arguments from the command line are available inside your program via the string slice os.Args, provided you have imported the package os. The flag package has a more sophisticated interface, and also provides a way to parse flags. Take this example from a DNS query tool:

dnssec := flag.Bool("dnssec", false, "Request DNSSEC records") 1
port := flag.String("port", "53", "Set the query port") 2
flag.Usage = func() {   3
    fmt.Fprintf(os.Stderr, "Usage: %s [OPTIONS] [name ...]\n", os.Args[0])
    flag.PrintDefaults() 4
}
flag.Parse() 5

At 1 we define a bool flag -dnssec. Note that this function returns a pointer to the value, the dnssec is now a pointer to a bool. At 2 we define an strings flag. Then at 3 we redefine the Usage variable of the flag package so we can add some extra text. The PrintDefaults at 4 5 will output the default help for the flags that are defined. Note even without redefining a flag.Usage the flag -h is supported and will just output the help text for each of the flags. Finally at 4 5 we call Parse that parses the command line and fills the variables.

After the flags have been parsed you can used them: if *dnssec { ... }

Executing commands

The os/exec package has functions to run external commands, and is the premier way to execute commands from within a Go program. It works by defining a *exec.Cmd structure for which it defines a number of methods. Let's execute ls -l:

import "os/exec"

cmd := exec.Command("/bin/ls", "-l")
err := cmd.Run()

The above example just runs "ls -l" without doing anything with the returned data, capturing the standard output from a command is done as follows:

cmd := exec.Command("/bin/ls", "-l")
buf, err := cmd.Output()

And buf is byte slice, that you can further use in your program.

Networking

All network related types and functions can be found in the package net. One of the most important functions in there is Dial. When you Dial into a remote system the function returns a Conn interface type, which can be used to send and receive information. The function Dial neatly abstracts away the network family and transport. So IPv4 or IPv6, TCP or UDP can all share a common interface.

Dialing a remote system (port 80) over TCP, then UDP and lastly TCP over IPv6 looks like this19:

conn, e := Dial("tcp", "192.0.32.10:80")
conn, e := Dial("udp", "192.0.32.10:80")
conn, e := Dial("tcp", "[2620:0:2d0:200::10]:80")

If there were no errors (returned in e), you can use conn to read and write. And conn implements the io.Reader and io.Writer interface. 20

But these are the low level nooks and crannies, you will almost always use higher level packages, such as the http package. For instance a simple Get for http:

package main

import (
    "fmt"
    "http"
    "io/ioutil"
)

func main() {
    r, err := http.Get("http://www.google.com/robots.txt")
    if err != nil {
        fmt.Printf("%s\n", err.String())
        return
    }
    b, err := ioutil.ReadAll(r.Body)
    r.Body.Close()
    if err == nil {
        fmt.Printf("%s", string(b))
    }
}

Exercises

Finger daemon

Write a finger daemon that works with the finger(1) command.

From the Debian package description:

Fingerd is a simple daemon based on RFC 1196 that provides an interface to the "finger" program at most network sites. The program is supposed to return a friendly, human-oriented status report on either the system at the moment or a particular person in depth.

Stick to the basics and only support a username argument. If the user has a .plan file show the contents of that file. So your program needs to be able to figure out:

Answer

package main

import (
	"bufio"
	"errors"
	"flag"
	"io/ioutil"
	"net"
	"os/user"
)

func main() {
	flag.Parse()
	ln, err := net.Listen("tcp", ":79")
	if err != nil {
		panic(err)
	}
	for {
		conn, err := ln.Accept()
		if err != nil {
			continue
		}
		go handleConnection(conn)
	}
}

func handleConnection(conn net.Conn) {
	defer conn.Close()
	reader := bufio.NewReader(conn)
	usr, _, _ := reader.ReadLine()

	if info, err := getUserInfo(string(usr)); err != nil {
		conn.Write([]byte(err.Error()))
	} else {
		conn.Write(info)
	}
}

func getUserInfo(usr string) ([]byte, error) {
	u, e := user.Lookup(usr)
	if e != nil {
		return nil, e
	}
	data, err := ioutil.ReadFile(u.HomeDir + ".plan")
	if err != nil {
		return data, errors.New("User doesn't have a .plan file!\n")
	}
	return data, nil
}

Echo server

Write a simple echo server. Make it listen to TCP port number 8053 on localhost. It should be able to read a line (up to the newline), echo back that line and then close the connection.

Make the server concurrent so that every request is taken care of in a separate goroutine.

Answer

A simple echo server might be:

package main

import (
	"bufio"
	"fmt"
	"net"
)

func main() {
	l, err := net.Listen("tcp", "127.0.0.1:8053")
	if err != nil {
		fmt.Printf("Failure to listen: %s\n", err.Error())
	}
	for {
		if c, err := l.Accept(); err == nil {
			Echo(c)
		}
	}
}

func Echo(c net.Conn) {
	defer c.Close()
	line, err := bufio.NewReader(c).ReadString('\n')
	if err != nil {
		fmt.Printf("Failure to read: %s\n", err.Error())
		return
	}
	_, err = c.Write([]byte(line))
	if err != nil {
		fmt.Printf("Failure to write: %s\n", err.Error())
		return
	}
}

When started you should see the following:

% nc 127.0.0.1 8053
Go is *awesome*
Go is *awesome*

To make the connection handling concurrent we only need to change one line in our echo server, the line:

if c, err := l.Accept(); err == nil { Echo(c) }

becomes:

if c, err := l.Accept(); err == nil { go Echo(c) }

Word and Letter Count

Write a small program that reads text from standard input and performs the following actions:

In other words implement wc(1) (check you local manual page), however you only have to read from standard input.

Answer

The following program is an implementation of wc(1).

package main

import (
	"bufio"
	"fmt"
	"os"
	"strings"
)

func main() {
	var chars, words, lines int
	r := bufio.NewReader(os.Stdin) 1
	for {
		switch s, ok := r.ReadString('\n'); true { 2
		case ok != nil: 3
			fmt.Printf("%d %d %d\n", chars, words, lines)
			return
		default: 4
			chars += len(s)
			words += len(strings.Fields(s))
			lines++
		}
	}
}

At 1 we create a new reader that reads from standard input, we then read from the input at 2. And at 3 we check the value of ok and if we received an error, we assume it was because of a EOF, So we print the current values;. Otherwise 4 we count the charaters, words and increment the number lines.

Uniq

Write a Go program that mimics the function of the Unix uniq command. This program should work as follows, given a list with the following items:

'a' 'b' 'a' 'a' 'a' 'c' 'd' 'e' 'f' 'g'

it should print only those items which don't have the same successor:

'a' 'b' 'a' 'c' 'd' 'e' 'f' 'g'

The next listing is a Perl implementation of the algorithm.

#!/usr/bin/perl
my @a = qw/a b a a a c d e f g/;
print my $first = shift @a; 
foreach (@a) {
    if ($first ne $_) { print; $first = $_; }
}

Answer

The following is a uniq implementation in Go.

package main

import "fmt"

func main() {
	list := []string{"a", "b", "a", "a", "c", "d", "e", "f"}
	first := list[0]

	fmt.Printf("%s ", first)
	for _, v := range list[1:] {
		if first != v {
			fmt.Printf("%s ", v)
			first = v
		}
	}
}

Quine

A Quine is a program that prints itself. Write a Quine in Go.

Answer

/* Go quine */
package main
import "fmt"
func main() {
 fmt.Printf("%s%c%s%c\n", q, 0x60, q, 0x60)
}
var q = `/* Go quine */
package main
import "fmt"
func main() {
 fmt.Printf("%s%c%s%c\n", q, 0x60, q, 0x60)
}
var q = `

Processes

Write a program that takes a list of all running processes and prints how many child processes each parent has spawned. The output should look like:

Pid 0 has 2 children: [1 2]
Pid 490 has 2 children: [1199 26524]
Pid 1824 has 1 child: [7293]

Here is a Perl version to help you on your way (or to create complete and utter confusion).

#!/usr/bin/perl -l
my (%child, $pid, $parent);
my @ps=`ps -e -opid,ppid,comm`;	  # capture the output from `ps`
foreach (@ps[1..$#ps]) {	  # discard the header line
  ($pid, $parent, undef) = split; # split the line, discard 'comm'
  push @{$child{$parent}}, $pid;  # save the child PIDs on a list
}
# Walk through the sorted PPIDs
foreach (sort { $a <=> $b } keys %child) {  
  print "Pid ", $_, " has ", @{$child{$_}}+0, " child",
    @{$child{$_}} == 1 ? ": " : "ren: ", "[@{$child{$_}}]";
}

Answer

There is lots of stuff to do here. We can divide our program up in the following sections:

In the solution presented below, we've used a map[int][]int, i.e. a map indexed with integers, pointing to a slice of ints -- which holds the PIDs. The builtin append is used to grow the integer slice.

A possible program is:

package main

import (
	"fmt"
	"os/exec"
	"sort"
	"strconv"
	"strings"
)

func main() {
	ps := exec.Command("ps", "-e", "-opid,ppid,comm")
	output, _ := ps.Output()
	child := make(map[int][]int)
	for i, s := range strings.Split(string(output), "\n") {
		if i == 0 { // kill first line
			continue
		}
		if len(s) == 0 { // kill last line
			continue
		}
		f := strings.Fields(s)
		fpp, _ := strconv.Atoi(f[1]) // parent's pid
		fp, _ := strconv.Atoi(f[0])  // child's pid
		child[fpp] = append(child[fpp], fp)
	}
	schild := make([]int, len(child))
	i := 0
	for k, _ := range child {
		schild[i] = k
		i++
	}
	sort.Ints(schild)
	for _, ppid := range schild {
		fmt.Printf("Pid %d has %d child", ppid, len(child[ppid]))
		if len(child[ppid]) == 1 {
			fmt.Printf(": %v\n", child[ppid])
			continue
		}
		fmt.Printf("ren: %v\n", child[ppid])
	}
}

Number cruncher

An example. We have picked the numbers: 1, 6, 7, 8, 8 and 75. And \(i\) is 977. This can be done in many different ways, one way is: \( ((((1 * 6) * 8) + 75) * 8) - 7 = 977\) or \( (8*(75+(8*6)))-(7/1) = 977\)

Implement a number cruncher that works like that. Make it print the solution in a similar format (i.e. output should be infix with parenthesis) as used above.

Calculate all possible solutions and show them (or only show how many there are). In the example above there are 544 ways to do it.

Answer

The following is one possibility. It uses recursion and backtracking to get an answer. When starting permrec we give 977 as the first argument:

% ./permrec 977
1+(((6+7)*75)+(8/8)) = 977  #1
...                         ...
((75+(8*6))*8)-7 = 977      #542
(((75+(8*6))*8)-7)*1 = 977  #543
(((75+(8*6))*8)-7)/1 = 977  #544

package main

import (
	"flag"
	"fmt"
	"strconv"
)

const (
	_ = 1000 * iota
	ADD
	SUB
	MUL
	DIV
	MAXPOS = 11
)

var mop = map[int]string{ADD: "+", SUB: "-", MUL: "*", DIV: "/"}
var (
	ok    bool
	value int
)

type Stack struct {
	i    int
	data [MAXPOS]int
}

func (s *Stack) Reset()     { s.i = 0 }
func (s *Stack) Len() int   { return s.i }
func (s *Stack) Push(k int) { s.data[s.i] = k; s.i++ }
func (s *Stack) Pop() int   { s.i--; return s.data[s.i] }

var found int
var stack = new(Stack)

func main() {
	flag.Parse()
	list := []int{1, 6, 7, 8, 8, 75, ADD, SUB, MUL, DIV}
	magic, ok := strconv.Atoi(flag.Arg(0)) // Arg0 is i
	if ok != nil {
		return
	}
	f := make([]int, MAXPOS)
	solve(f, list, 0, magic)
}

func solve(form, numberop []int, index, magic int) {
	var tmp int
	for i, v := range numberop {
		if v == 0 {
			goto NEXT
		}
		if v < ADD { // it's a number, save it
			tmp = numberop[i]
			numberop[i] = 0
		}
		form[index] = v
		value, ok = rpncalc(form[0 : index+1])

		if ok && value == magic {
			if v < ADD {
				numberop[i] = tmp // reset and go on
			}
			found++
			fmt.Printf("%s = %d  #%d\n", rpnstr(form[0:index+1]), value, found)
		}

		if index == MAXPOS-1 {
			if v < ADD {
				numberop[i] = tmp // reset and go on
			}
			goto NEXT
		}
		solve(form, numberop, index+1, magic)
		if v < ADD {
			numberop[i] = tmp // reset and go on
		}
	NEXT:
	}
}

func rpnstr(r []int) (ret string) { // Convert rpn to infix notation
	s := make([]string, 0) // Still memory intensive
	for k, t := range r {
		switch t {
		case ADD, SUB, MUL, DIV:
			var a, b string
			a, s = s[len(s)-1], s[:len(s)-1]
			b, s = s[len(s)-1], s[:len(s)-1]
			if k == len(r)-1 {
				s = append(s, b+mop[t]+a)
			} else {
				s = append(s, "("+b+mop[t]+a+")")
			}
		default:
			s = append(s, strconv.Itoa(t))
		}
	}
	for _, v := range s {
		ret += v
	}
	return
}

func rpncalc(r []int) (int, bool) {
	stack.Reset()
	for _, t := range r {
		switch t {
		case ADD, SUB, MUL, DIV:
			if stack.Len() < 2 {
				return 0, false
			}
			a := stack.Pop()
			b := stack.Pop()
			if t == ADD {
				stack.Push(b + a)
			}
			if t == SUB {
				// disallow negative subresults
				if b-a < 0 {
					return 0, false
				}
				stack.Push(b - a)
			}
			if t == MUL {
				stack.Push(b * a)
			}
			if t == DIV {
				if a == 0 {
					return 0, false
				}
				// disallow fractions
				if b%a != 0 {
					return 0, false
				}
				stack.Push(b / a)
			}
		default:
			stack.Push(t)
		}
	}
	if stack.Len() == 1 { // there is only one!
		return stack.Pop(), true
	}
	return 0, false
}

Bibliography

  1. C. A. R. Hoare. Quicksort. http://en.wikipedia.org/wiki/Quicksort 1960.
  2. Rob Pike. The Go programming language, day 3. http://golang.org/doc/GoCourseDay3.pdf 2010.
  3. James Gosling et al. Java. http://oracle.com/java/ 1995.
  4. Perl Package Authors. Comprehensive perl archive network. http://cpan.org/ 2010.
  5. Plan 9 Authors. Plan 9. http://plan9.bell-labs.com/plan9/index.html 1992.
  6. Ericsson Cooperation. Erlang. http://www.erlang.se/ 1986.
  7. Larry Wall et al. Perl. http://perl.org/ 1987.
  8. Go Authors. Go youtube channel. http://youtube.com/gocoding/ 2010.
  9. Go Authors. Go tutorial. http://golang.org/doc/go_tutorial.html 2010.
  10. Imran On Tech. Using fizzbuzz to find developers.... 2010.
  11. Go Community. Function accepting a slice of interface types. 2010.
  12. D. Zimmerman. The finger user information protocol. http://www.ietf.org/rfc/rfc1196.txt 1990.
  13. Go Authors. Go website. http://golang.org/ 2010.
  14. Go Authors. Defer, panic, and recover. https://blog.golang.org/defer-panic-and-recover 2010.
  15. Ian Lance Taylor. Go interfaces. http://www.airs.com/blog/archives/277 2010.
  16. Inferno Authors. Inferno. http://www.vitanuova.com/inferno/ 1995.
  17. Go Authors. Go 1.5 Release Notes. https://golang.org/doc/go1.5 2010.
  18. Mark C. Chu-Carroll. Google's new language: Go. http://scienceblogs.com/goodmath/2009/11/googles_new_language_go.php 2010.
  19. LAMP Group at EPFL. Scala. 2003.
  20. Go Authors. Go faq. http://golang.org/doc/go_faq.html 2010.
  21. Go Authors. Effective Go. http://golang.org/doc/effective_go.html 2010.
  22. Haskell Authors. Haskell. http://www.haskell.org/ 1990.
  23. Rob Pike. Newsqueak: a language for communicating with mice. http://swtch.com/~rsc/thread/newsqueak.pdf 1989.
  24. Bjarne Stroustrup. The C++ programming language. 1983.
  25. Go Authors. Go release history. http://golang.org/doc/devel/release.html 2010.
  26. Wikipedia. Iota. http://en.wikipedia.org/wiki/Iota 2010.
  27. D. Crockford. The application/json media type for javascript object notation (json). http://www.ietf.org/rfc/rfc4627.txt 2006.
  28. Go Authors. Getting Started. http://golang.org/doc/install/ 2012.
  29. Go Authors. Go package documentation. http://golang.org/doc/pkg/ 2010.
  30. Brian Kernighan Dennis Ritchie. The C programming language. 1975.
  31. Go Community. Go issue 65: compiler can't spot guaranteed return in if statement. 2010.
  32. Kolkman & Gieben. Dnssec operational practices. http://www.ietf.org/rfc/rfc4641.txt 2006.
  33. Wikipedia. Bubble sort. http://en.wikipedia.org/wiki/Bubble_sort 2010.
  34. Go Community. Go Nuts mailing list. http://groups.google.com/group/golang-nuts 2010.
  35. Go Authors. Go for C++ programmers. http://golang.org/doc/go_for_cpp_programmers.html 2010.
  36. Go Authors. The Go programming language blog. http://blog.golang.org/ 2010.
  37. Rob Pike. The Go programming language, day 2. http://golang.org/doc/GoCourseDay2.pdf 2010.
  38. Russ Cox. Profiling Go programs. http://blog.golang.org/2011/06/profiling-go-programs.html 2011.
  39. C. A. R. Hoare. Communicating sequential processes (csp). http://www.usingcsp.com/cspbook.pdf 1985.
  40. Go Authors. Go language specification. http://golang.org/doc/go_spec.html 2010.
  41. Wikipedia. Duck typing. http://en.wikipedia.org/wiki/Duck_typing 2010.
  42. Plan 9 Authors. Limbo. http://www.vitanuova.com/inferno/papers/limbo.html 1995.

Footnotes

  1. Yes, that sounds a lot like coroutines, but goroutines are slightly different as we will see in .
  2. http://play.golang.org.
  3. http://golang.org/doc/ itself is served by godoc.
  4. When building from source it must be installed separately with go get golang.org/x/tools/cmd/godoc.
  5. The word iota is used in a common English phrase, 'not one iota', meaning 'not the slightest difference', in reference to a phrase in the New Testament: "until heaven and earth pass away, not an iota, not a dot, will pass from the Law."
  6. In the UTF-8 world characters are sometimes called runes .
  7. You can use the command godoc builtin to read the online documentation about the built-in types and functions.
  8. Always rainy in March anyway.
  9. This is a motto of Go; "Do more with less code".
  10. A function literal is sometimes called a closure .
  11. Modified from a presentation of Eleanor McHugh.
  12. The descriptions are copied from the packages' go doc.
  13. The downside is that you know have to worry about garbage collection. If you really need it garbage collection in a Go program can be disabled by running it with the environment variable GOGC set to off: GOGC=off ./myprogram.
  14. Taking the address of a composite literal tells the compiler to allocate it on the heap, not the stack.
  15. Also see in this chapter.
  16. The following text is partly from .
  17. Text copied from .
  18. go doc reflect
  19. In case you are wondering, 192.0.32.10 and 2620:0:2d0:200::10 are http://www.example.org.
  20. The variable conn also implements a close method, this really makes it an io.ReadWriteCloser.

Index

a

array multidimensional 1

b

built-in recover 1 panic 1 new 1,2 make 1,2 close 1 imag 1 copy 1,2 complex 1 cap 1 len 1 delete 1 print 1 println 1 real 1 append 1,2

c

channel blocking read 1 unbuffered 1 non-blocking write 1 blocking write 1 non-blocking read 1 channels 1,2 closure 1,2 complex numbers 1

d

duck typing 1

f

field anonymous 1 1 functions receiver 1 method 1 literal 1 as values 1 exported 1 pass-by-value 1 variadic 1 private 1 public 1 named return parameters 1 literals 1 signature 1

g

generic 1 goroutine 1,2

i

interface 1 set of methods 1 type 1 value 1 io buffered 1 io.Reader 1

k

keywords map 1 import 1 map remove elements 1 defer list 1 else 1 break 1 for 1 continue 1 select 1 goto 1 range 1,2,3,4 struct 1 go 1 fallthrough 1 switch 1 map adding elements 1 type 1 map existence 1 package 1 return 1 if 1 defer 1 iota 1 default 1

l

label 1 literal composite 1,2

m

methods inherited 1

n

networking Dial 1 nil 1

o

operators increment 1 channel 1 and 1 address-of 1 or 1 bit wise xor 1 bitwise or 1 bitwise and 1 bitwise clear 1 not 1

p

package strconv 1 flag 1,2 fmt 1,2 os 1 reflect 1,2 bytes 1 bufio 1,2,3 sync 1 net/http 1 sort 1 unsafe 1 os/exec 1,2 encoding/json 1 builtin 1 html/template 1 ring 1 io 1,2

r

reference types 1 runes 1,2,3

s

scope local 1,2 slice length 1 capacity 1 structures embed 1

t

tooling go test 1 go build 1 go run 1 go 1 type assertion 1 type switch 1

v

variables underscore 1 declaring 1 parallel assignment 1 assigning 1