CMake is a set of open-source, cross-platform tools designed to build, test and package software. It uses compiler independent configuration files to generate native makefiles and project configurations that can be used with many compiler environments and IDEs such as make, Xcode or Visual Studio.

In this blog article I want to introduce the basic usage of CMake. I will show how to use CMake to build a simple command line executable with make and Xcode. Furthermore, I’m going to demonstrate CMakes commands to generate files from template files.

The project I use for the demonstration consits of two files. First, there is a template for a config header file. The template file is called CMake will be used to generate the file config.h from the template file.

#include <string>

const std::string Hello = "@HELLO@";
const std::string World = "@WORLD@";

As you can see there are two constants defined in Hello and World. The strings @HELLO@ and @WORLD@ are variables defined in the file CMakeLists.txt (see below). The two constants are used in the second file called main.cpp.

#include "config.h"
#include <iostream>

int main() {
    std::cout << Hello << ", " << World << "!\n";

Last but not least, CMake needs a file called CMakeLists.txt in the directory. It contains all the CMake commands.

cmake_minimum_required (VERSION 3.9)
project (hello)

set (HELLO Hello)
set (WORLD World)

configure_file (

add_executable(hello main.cpp)

target_include_directories(hello PRIVATE "${PROJECT_BINARY_DIR}")

This file first sets the minimum required version of the CMake package using the command cmake_minimum_required(). In this case the version 3.9 or newer is required. The command project() is used to set the name of the entire project. Before the file config.h can be generated the variables HELLO and WORLD have to be defined. This is done with the set() command. Subsequently, the configure_file() command copies the file to config.h and replaces the strings @HELLO@ and @WORLD@ with the values of the defined variables. The command add_executable() is used to define the resulting target executable using the main.cpp. Because the generated file config.h is created in the binary directory the path to the binary directory needs to be added to the INCLUDE_DIRECTORIES. This is done with the help of the target_include_directories() command.

There are many more CMake commands available. Execute man cmake-commands or visit the CMake website for a complete reference of all commands.

Two steps need to be done to build the executable. First, the tool cmake has to be executed on the directory containing the source files, namely main.cpp,, and CMakeLists.txt. When executed without any options cmake will create a Makefile and a few other files.

To separate the generated files from the source code it is helpful to create a build directory. Builds like this are called out-of-source builds and they makes it easy to have builds for different platforms and helps to avoid unintentionally commiting generated files to the source control.

$ mkdir build
$ cd build
$ cmake ..

At this point the second step can be done by running make to build the executable.

$ make

Now the directory contains the executable file hello which can be executed.

$ ./hello
Hello, World!

As mentioned above CMake can generate project files for many build systems and IDEs. The command line option -G specifies the generator. For Mac it can create project files for Xcode.

$ mkdir build-xcode
$ cd build-xcode
$ cmake -G Xcode ..

This command creates the Xcode project hello.xcodeproj. This project directory can be opened in Xcode or xcodebuild can be used to build the executable on the command line as follows.

$ xcodebuild -project hello.xcodeproj -alltargets -configuration Release

In this case the executable is created in the Release directory.

$ Release/hello
Hello, World!