How To Build Go Executables for Multiple Platforms on Ubuntu 20.04

Introduction

The Go programming language comes with a rich toolchain that makes obtaining packages and building executables incredibly easy. One of Go’s most powerful features is the ability to cross-build executables for any Go-supported foreign platform. This makes testing and package distribution much easier, because you don’t need to have access to a specific platform in order to distribute your package for it.

In this tutorial, you’ll use Go’s tools to obtain a package from version control and automatically install its executable. Then you’ll manually build and install the executable so you can be familiar with the process. Then you’ll build an executable for a different architecture, and automate the build process to create executables for multiple platforms. When you’re done, you’ll know how to build executables for Windows and macOS, as well as other platforms you want to support.

Prerequisites

To follow this tutorial, you will need:

• One Ubuntu 20.04 server set up by following the Ubuntu 20.04 initial server setup guide, including a sudo non-root user and a firewall.
• Go installed, as described in How to Install Go on Ubuntu 20.04.

Step 1 — Creating a simple Go program

Now that Go is installed, you can try creating your Hello, World!.

First, create a new directory for your Go workspace, which is where Go will build its files:

mkdir hello

Then move into the directory you just created:

cd hello

When importing packages, you have to manage the dependencies through the code’s own module. You can do this by creating a go.mod file with the go mod init command:

go mod init hello

Next, create a Hello, World! Go file in your preferred text editor:

nano hello.go

Add the following text into your hello.go file:

package main

import "fmt"

func main() {
    fmt.Println("Hello, World!")
}

Then, save and close the file by pressing CTRL+X, then Y, and then ENTER.

Test your code to check that it prints the Hello, World! greeting:

go run .

Output
Hello, World!

The go run command compiles and runs the Go package from a list of .go source files from the new hello directory you created and the path you imported. But, you can also use go build to make an executable file that can save you some time.

Step 2 — Building an Executable

The go run command ran the code for your “Hello, World!” program, but you may want to build your program into a binary to run anywhere on your system, not just from the source. The go build command builds executables.

go build hello

As before, no output indicates successful operation. The executable will be generated in your current directory, with the same name as the directory containing the package. In this case, the executable will be named hello.

If you’re located in the package directory, you can omit the path to the package and simply run go build.

To specify a different name or location for the executable, use the -o flag. Let’s build an executable called hello and place it in a build directory within the current working directory:

go build -o build/hello hello

This command creates the executable, and also creates the ./build directory if it doesn’t exist.

Now let’s look at installing executables.

Step 3 — Installing an Executable

Building an executable creates the executable file in the current directory or the directory of your choice. Installing an executable is the process of creating an executable and storing it in $GOPATH/bin. The go install command works just like go build, but go install takes care of placing the output file in the right place for you.

To install an executable, use go install, followed by the package import path. Once again, use your “Hello, World!” program to try this out:

go install hello

Just like with go build, you’ll see no output if the command was successful. And like before, the executable is created with the same name as the directory containing the package. But this time, the executable is stored in $GOPATH/bin. If $GOPATH/bin is part of your $PATH environment variable, the executable will be available from anywhere on your operating system. You can verify its location using which command:

which hello

You’ll see the following output:

Output of which
/home/sammy/go/bin/hello

Now that you understand how go get, go build, and go install work, and how they’re related, let’s explore one of the most popular Go features: creating executables for other target platforms.

Step 4 — Building Executables for Different Architectures

The go build command lets you build an executable file for any Go-supported target platform, on your platform. This means you can test, release and distribute your application without building those executables on the target platforms you wish to use.

Cross-compiling works by setting required environment variables that specify the target operating system and architecture. We use the variable GOOS for the target operating system, and GOARCH for the target architecture. To build an executable, the command would take this form:

GOOS=target-OS GOARCH=target-architecture go build package-import-path

The OS and architecture are set before running the go build command. This lets you use environment variables for the current command execution only. The variables are unset or reset after the command executes.

In order to find out what operating systems and platforms are available for building executables, you can use the dist tool:

go tool dist list

This will provide you a list of operating systems and architectures separated by / characters:

Output
aix/ppc64
android/386
android/amd64
android/arm
android/arm64
darwin/amd64
darwin/arm64
dragonfly/amd64
freebsd/386
freebsd/amd64
freebsd/arm
freebsd/arm64
illumos/amd64
ios/amd64
ios/arm64
js/wasm
linux/386
linux/amd64
linux/arm
linux/arm64
linux/mips
linux/mips64
linux/mips64le
linux/mipsle
linux/ppc64
linux/ppc64le
linux/riscv64
linux/s390x
netbsd/386
netbsd/amd64
netbsd/arm
netbsd/arm64
openbsd/386
openbsd/amd64
openbsd/arm
openbsd/arm64
openbsd/mips64
plan9/386
plan9/amd64
plan9/arm
solaris/amd64
windows/386
windows/amd64
windows/arm

Using the values in the list, we can build hello for Windows 64-bit like this:

GOOS=windows GOARCH=amd64 go build hello

Once again, no output indicates that the operation was successful. The executable will be created in the current directory, using the package name as its name. However, since we built this executable for Windows, the name ends with the suffix .exe.

You should have a hello.exe file in your current directory, which you can verify with the ls command.

ls hello.exe

You’ll see the hello.exe file listed in the output:

Output
hello.exe

Let’s look at scripting this process to make it easier to release software for multiple target environments.

Step 5 — Creating a Script to Automate Cross-Compilation

The process of creating executables for many platforms can be a little tedious, but we can create a script to make things easier.

The script will take the package import path as an argument, iterate through a predefined list of operating system and platform pairs, and generate an executable for each pair, placing the output in the current directory. Each executable will be named with the package name, followed by the target platform and architecture, in the form package-OS-architecture. This will be a universal script that you can use on any project.

Within your project’s directory, create a new file called go-executable-build.bash in your text editor:

nano go-executable-build.bash

We will start our script with a shebang line. This line defines which interpreter will parse this script when it’s run as an executable. Add the following line to specify that bash should execute this script:

#!/usr/bin/env bash

We want to take the package import path as a command line argument. To do this, we will use the $n variable, where n is a non-negative number. The variable $0 contains the name of the script you executed, while $1 and greater will contain user-provided arguments. Add this line to the script, which will take the first argument from the command line and store it in a variable called package:

...
package=$1

Next, make sure that the user provided this value. If the value isn’t provided, exit the script with a message explaining how to use the script:

...

if [[ -z "$package" ]]; then
  echo "usage: $0 <package-name>"
  exit 1
fi

This if statement checks the value of the $package variable. If it’s not set, we use echo to print the correct usage, and then terminate the script using exit. exit takes a return value as an argument, which should be 0 for successful executions and any non-zero value for unsuccessful executions. We use 1 here since the script wasn’t successful.

...
package="github.com/user/hello"

Next, we want to extract the package name from the path. The package import path is delimited by / characters, with the package name located at the end of the path. First, we will split the package import path into an array, using the / as the delimiter:

package_split=(${package//\// })

The package name should be the last element of this new $package_split array. In Bash, you can use a negative array index to access an array from the end instead of the beginning. Add this line to grab the package name from the array and store it in a variable called package_name:

...
package_name=${package_split[-1]}

Now, you’ll need to decide what platforms and architectures you want to build executables for. In this tutorial, we’ll build executables for Windows 64-bit, Windows 32-bit, and 64-bit macOS. We will put these targets in an array with the format OS/Platform, so we can split each pair into GOOS and GOARCH variables using the same method we used to extract the package name from the path. Add the platforms to the script:

...
platforms=("windows/amd64" "windows/386" "darwin/amd64")

Next, we’ll iterate through the array of platforms, split each platform entry into values for the GOOS and GOARCH environment variables, and use those to build the executable. We can do that with the following for loop:

...
for platform in "${platforms[@]}"
do
    ...
done

The platform variable will contain an entry from the platforms array in each iteration. We need to split platform into two variables - GOOS and GOARCH. Add the following lines to the for loop:

for platform in "${platforms[@]}"
do
    platform_split=(${platform//\// })
    GOOS=${platform_split[0]}
    GOARCH=${platform_split[1]}
    
done

Next, we will generate the executable’s name by combining the package name with the OS and architecture. When we’re building for Windows, we also need to add the .exe suffix to the filename. Add this code to the for loop:

for platform in "${platforms[@]}"
do
    platform_split=(${platform//\// })
    GOOS=${platform_split[0]}
    GOARCH=${platform_split[1]}
    
    output_name=$package_name'-'$GOOS'-'$GOARCH

    if [ $GOOS = "windows" ]; then
        output_name+='.exe'
    fi
done

With the variables set, we use go build to create the executable. Add this line to the body of the for loop, right above the done keyword:

...
    if [ $GOOS = "windows" ]; then
        output_name+='.exe'
    fi
    
    env GOOS=$GOOS GOARCH=$GOARCH go build -o $output_name $package

done

Finally, we should check to see if there were errors building the executable. For example, we might run into an error if we try to build a package we don’t have sources for. We can check the go build command’s return code for a non-zero value. The variable $? contains the return code from a previous command’s execution. If go build returns anything other than 0, there is a problem, and we’ll want to exit the script. Add this code to the for loop, after the go build command and above the done keyword.

...

    env GOOS=$GOOS GOARCH=$GOARCH go build -o $output_name $package

    if [ $? -ne 0 ]; then
           echo 'An error has occurred! Aborting the script execution...'
        exit 1
    fi

With that, we now have a script that will build multiple executables from our Go package. Here’s the completed script:

#!/usr/bin/env bash

package=$1
if [[ -z "$package" ]]; then
  echo "usage: $0 <package-name>"
  exit 1
fi
package_split=(${package//\// })
package_name=${package_split[-1]}
    
platforms=("windows/amd64" "windows/386" "darwin/amd64")

for platform in "${platforms[@]}"
do
    platform_split=(${platform//\// })
    GOOS=${platform_split[0]}
    GOARCH=${platform_split[1]}
    output_name=$package_name'-'$GOOS'-'$GOARCH
    if [ $GOOS = "windows" ]; then
        output_name+='.exe'
    fi    

    env GOOS=$GOOS GOARCH=$GOARCH go build -o $output_name $package
    if [ $? -ne 0 ]; then
           echo 'An error has occurred! Aborting the script execution...'
        exit 1
    fi
done

Verify that your file matches the preceding code. Then save the file and exit the editor.

Before we can use the script, we have to make it executable with the chmod command:

chmod +x go-executable-build.bash

Finally, test the script by building executables for hello:

./go-executable-build.bash hello

If everything goes well, you should have executables in your current directory. No output indicates successful script execution. You can verify are executables created using ls command:

ls hello*

You should see all three versions:

Example ls output
hello-darwin-amd64 hello-windows-386.exe hello-windows-amd64.exe

To change the target platforms, just change the platforms variable in your script.

Conclusion

In this tutorial, you’ve learned how to use Go’s tooling to obtain packages from version control systems, as well as build and cross-compile executables for different platforms.

You also created a script that you can use to cross-compile a single package for many platforms.

To make sure your application works correctly, you can take a look at testing and continuous integration like Travis-CI and AppVeyor for testing on Windows.