An introduction to the build pipeline, and how we can orchestrate it manually

Before working with CMake, it's useful to understand the tools it orchestrates. This lesson guides you through setting up a C++ toolchain.

Building C++ Programs

Walking through the steps of the C++ compilation process, from source code to machine code.

The Compilation Pipeline

The linker's role in the build process, how it resolves symbols to combine object files, and how to troubleshoot common errors.

Object Files and the Linking Process

The difference between static and dynamic libraries, how to create them, and the trade-offs between them.

Static and Shared Libraries

The C++ Build Process

Recommended approaches for managing large projects and the libraries that they depend on

Learn how to structure your C++ projects for scalability and maintainability, from basic directory layouts to managing multiple modules.

Organizing a C++ Project

Learn to manage third-party libraries in C++, covering include/library paths, static vs. shared libraries, versioning, and platform differences.

Managing Libraries and Dependencies

An introduction to traditional build tools like Makefiles and IDE projects, highlighting the maintenance, scalability, and cross-platform challenges that result

Automated Build Systems and Their Limitations

Project Structure and Dependency Management

Setting up CMake and creating your first project

An introduction to CMake, the cross-platform, open-source meta-build system that solves the core challenges of C++ project management.

Why CMake?

A step-by-step guide to installing CMake. We'll explore command-line, GUI, and IDE workflows, and how to configure your compiler.

Setting Up a CMake Environment

Creating the bare minimum `CMakeLists.txt` file and build up to a project with an executable and a library, learning the fundamental commands along the way.

Writing a CMakeLists File

Learn the two-stage process of building a CMake project. This lesson covers configuring, generating, building, running, and troubleshooting common errors.

Building and Running CMake Projects

Learn how to organize large C++ projects in CMake using subdirectories and the `add_subdirectory()` command to create modular, maintainable builds.

CMake Project Structure and Subdirectories

Introducing CMake

Exploring CMake's programming features, caching mechanisms, and interactions with the file system

A primer on the fundamental building blocks of the CMake language: variables, string interpolation, and logging messages.

CMake Variables and Logging

Discover the CMake Cache, the mechanism for storing persistent user-configurable options, and learn how to create and manage cached variables.

The CMake Cache

Exploring CMake's command syntax and its list data type, including how to create, manipulate, and use lists in your projects.

CMake Commands, Arguments, and Lists

Transform your static build descriptions into dynamic scripts with CMake's control flow commands, if() and foreach().

CMake Conditionals and Loops

Exploring CMake's `file()` command, covering how to read, write, copy, and glob files, and how to manipulate paths.

Working with the File System

Learn to create reusable logic with CMake functions and macros, and how to organize them into modules for clean, scalable build systems.

Modularizing CMake Code

CMake Language and Fundamentals

Learn more about targets, and how their properties and links are the foundation of modern CMake

Learn to work with target properties - understanding their purpose, setting and updating them, and retrieving them when your build requires it

Target Properties

Learn about target types, and how the `PUBLIC`, `PRIVATE`, and `INTERFACE` keywords control how properties are shared.

Relationships Between Targets

Discover how CMake's `target_link_libraries()` command builds a dependency graph, and how it uses this graph to automatically manage transitive dependencies and link order.

Indirect Relationships and the Dependency Graph

Learn to use abstract target types like `INTERFACE`, `ALIAS`, and `IMPORTED` to model complex project needs, organize build properties, and integrate pre-compiled binaries.

Using `INTERFACE`, `ALIAS`, and `IMPORTED` Libraries

Adding support for user-configurable library types and an initial introduction to target installation.

Using Shared Libraries

Targets and Dependencies

Learn how to manage multiple build configurations within the same project, and how to create scripts that adapt to the environment

Learn how CMake manages different build configurations like Debug and Release, the difference between generator types, and how to apply settings conditionally using modern techniques.

Build Configurations (Debug, Release, etc.)

Learn the tools and patterns needed to write portable CMakeLists.txt files that work seamlessly across platforms and compilers

Cross-Platform Configurations

Learn to write architecture-aware CMake scripts for 32/64-bit systems and Apple's Universal Binaries (Intel/ARM).

Handling Different Architectures

Learn to build for different operating systems and architectures using CMake's cross-compilation support and toolchain files.

Cross-Compilation and Toolchain Files

Platforms and Build Configurations

Learn how to add external libraries to our build, and how to create our own packages for others to use

Learn to find and use pre-installed libraries with `find_package()`, understanding Module vs. Config mode and the power of imported targets

Finding Dependencies using `find_package()`

Learn how to make your libraries consumable by other projects using CMake by turning your build-tree into a distributable install-tree.

Creating a Consumable Package

Learn how to consume a CMake package you've installed, completing the producer-consumer cycle. We'll cover using `find_package()` and how to point CMake to your library's location

Using Installed Packages

Learn how to create self-contained, reproducible builds by fetching dependencies from source using `FetchContent()` and `ExternalProject_Add()`.

Fetching External Dependencies

Creating and Consuming Packages

Integrating the vcpkg and Conan package managers to handle external dependencies

Learn to manage C++ dependencies effortlessly with the vcpkg package manager and its seamless CMake integration using toolchain files and manifests.

Integrating CMake and vcpkg

Learn the day-to-day commands for managing dependencies with vcpkg and how to guarantee reproducible builds using baselines and version overrides

vcpkg Commands and Versioning

Combine vcpkg with CMake's cross-compilation capabilities. Learn to use triplets to target different architectures and chain-load toolchain files

Cross-Compiling with vcpkg

Discover Conan, the binary-first C++ package manager, and learn how to integrate it with CMake using profiles and lockfiles for fast, reproducible builds.

Using Conan with CMake

Learn to cross-compile C++ projects with Conan, using build and host profiles to manage dependencies for different platforms.

Cross-Compilation with Conan

Package Managers

Introducing presets, allowing us to simplify our workflow and share common configurations across our team

Learn the fundamentals of CMake Presets. This lesson introduces `CMakePresets.json` to simplify complex build commands.

CMake Presets

Learn to structure your CMake presets with inheritance and customize them for your local environment using `CMakeUserPresets.json`.

User Presets and Inheritance

Learn to make your CMake presets portable by using environment variables and conditional logic to adapt to different machine configurations.

Environment Variables and Conditional Presets

Learn how to extend the build process beyond simple compilation by integrating external tools and custom commands

Learn how to use generator expressions `$<...>` for build-time conditional logic.

Generator Expressions and Conditional Logic

Learn how to attach commands to specific points in a target's build lifecycle using `add_custom_command(TARGET)` for tasks like post-build file copying.

Integrating Tools with `add_custom_command()`

Learn to integrate code generators like Protobuf into your build process using `add_custom_command()`, creating automated file-level dependencies in CMake.

Generating Files with Protobuf and `add_custom_command()`

Custom Commands

An introduction to automating common workflow chores by creating custom targets

Learn to create standalone, non-code targets using add_custom_target() for utility tasks like code formatting and documentation generation.

Custom Targets and Build Utilities

Using the Clang-Format tool to automatically enforce a consistent coding style, and integrating it into our build using a custom CMake target

Using Clang-Format

An introduction to file sets and the `target_sources()` command, allowing us to manage our header files in a more controlled way

Using File Sets

A step-by-step guide to integrating Doxygen into your CMake build process for automated C++ documentation.

Generating Documentation with Doxygen

Custom Targets

Integrate automated tests into a CMake project using CTest, GoogleTest, and GoogleMock

Learn the fundamentals of automated testing in C++ with CTest. Create your first test suite, register tests with `add_test()`, and run them with the `ctest` command.

Automated Testing and CTest

Learn to refactor manual C++ tests into a professional test suite using the GoogleTest framework, managed by vcpkg and integrated with CTest.

Integrating GoogleTest with CMake

Learn to use a wide range of assertions, write data-driven parameterized tests, and isolate dependencies with Google Mock.

Assertions and Parameterized Tests

Learn how to isolate dependencies that our project uses, and then how to replace those dependencies at test time using Google Mock.

Dependency Injection and Mocking

Learn the standard pattern for testing application logic by refactoring it into a linkable library

Testing Executables

Move beyond individual tests to manage a full test suite. Learn to organize tests, run tests in parallel, filter by labels, and measure our code coverage

Managing a Test Suite

Automated Testing

Using static and dynamic analysis in CMake projects to automatically detect memory bugs, performance issues, and problematic code

Learn to find memory bugs at runtime by integrating AddressSanitizer (ASan) into a CMake project

Using AddressSanitizer (ASan)

Adding benchmarks to our build to measure the performance of our code, and track how it changes over time

Using Google Benchmark

Integrating Clang-Tidy to enforce good practices and find bugs before we even compile

Using Clang-Tidy

Automated Analysis

Automating the entire development lifecycle using Git for souce control, GitHub Actions for cloud builds, and CPack to create distributable packages and installers.

An introduction to Git, the version control system that underpins modern collaborative software development and automated build pipelines.

Source Control with Git

Learn how to automate your build and test process with Continuous Integration (CI) and a build server. This lesson provides a guide to creating a GitHub Actions workflow for a CMake project.

Build Servers and Continuous Integration

Learn to package your C++ projects using CMake's CPack. This lesson covers creating developer ZIP archives and professional Windows MSI installers with components.

Packaging with CPack

Learn to automate the packaging process by integrating CPack with a multi-platform GitHub Actions workflow

Packaging with GitHub Actions

Learn to automate the creation of a professional Windows installer by integrating CPack, the WiX Toolset, and a multi-platform GitHub Actions workflow.

Creating Installers with CPack and GitHub Actions

Learn to automate the release process using a GitHub Actions workflow that builds, packages, and publishes artifacts to a GitHub Release

Deployment with GitHub Releases

Continuous Integration and Packaging

If you have a basic grasp of C++ but find the build process intimidating, this course is for you. We assume you know the language, but not how to manage large projects.

Manage complex, cross-platform builds using industry-standard tools. Create scalable build systems that automate the entire development lifecycle.

This course is a comprehensive deep-dive into the entire C++ build ecosystem, centered around the power and flexibility of modern CMake. We leave no stone unturned.

Start with a foundational understanding of compilers, linkers, static vs. shared libraries, and object files.

The Full Build Pipeline

Learn every aspect of the CMake language, from variables and control flow to functions, modules, and generator expressions.

Complete CMake Language Mastery

Master the modern target-centric paradigm, including properties, usage requirements, and abstract targets.

Advanced Target Management

Explore every method for handling dependencies, from `find_package` and `FetchContent` to vcpkg and Conan, including versioning and cross-compilation.

In-Depth Dependency Management

Integrate a full suite of professional tools for testing (CTest, GoogleTest), analysis (ASan, Clang-Tidy), and packaging (CPack).

Total Workflow Automation

Build a fully automated deployment pipeline that takes your code from a Git push to a published GitHub Release with multi-platform installers.

End-to-End CI/CD and Deployment

What You'll Learn

Become a software engineer with C++.  Starting from the basics, we guide you step by step along the way

Intro to C++ Programming

If you're completely new to programming or want a thorough introduction, consider starting with our free introductory course:

New to Programming?

Managing C++ Projects Using CMake

We've learned how to write intelligent build scripts using CMake's control flow commands like `if()` and `foreach()`. These commands allow us to make decisions and automate repetitive tasks during the **configure** phase of our build.


However, some decisions can't be made during configuration. As we saw in the <InlineLessonPromo url="cmake/cmake-build-configurations" title="Build Configurations (Debug, Release, etc.)" description="Learn how CMake manages different build configurations like Debug and Release, the difference between generator types, and how to apply settings conditionally using modern techniques.">previous lesson</InlineLessonPromo>, the choice between a `Debug` and `Release` build happens at a different time depending on the build system we generate.


This creates a portability problem. How can we write a single, clean `CMakeLists.txt` that applies settings conditionally, regardless of whether the user is building with Makefiles or Visual Studio?  As we saw earlier, the answer is a **generator expressions**.  In this lesson, we'll cover them in more depth.


## Introducing Generator Expressions


A **generator expression** is a special piece of syntax, written as `$<...>`, that solves this problem.


When CMake's parser encounters a generator expression, it does _**not**_ evaluate it. Instead, it copies the expression verbatim into the appropriate property field in the generated `Makefile` or `.vcxproj` file.


It is then the job of the native build tool (`make`, `MSBuild`, etc.) to evaluate this expression when you actually build your target. This makes the decision "just in time," using the build configuration active at that moment.


It's like a deferred function call. You tell CMake, "When you eventually build this, please evaluate this expression and use the result."


## Common Generator Expressions


While there are many types of generator expressions, a few are fundamental for managing configurations and platforms.


### The `CONFIG` Expression


As we saw earlier, this is the primary tool for build-type-specific logic. It compares the name of the active configuration to the string we provide.  It returns `"1"` (a true value) if they match and `"0"` (false) otherwise.


In the following example, we define the `IS_DEBUG_BUILD` preprocessor macro, but only for Debug builds:


```c++
#<cmake>
target_compile_definitions(GreeterApp PRIVATE
  $<$<CONFIG:Debug>:IS_DEBUG_BUILD>
)
```


The syntax can look a little strange at first because it's often nested.


`$<CONFIG:Debug>`: At build time, this inner expression is evaluated first. If the current configuration is `Debug`, it evaluates to the string `"1"`. Otherwise, it evaluates to `"0"`.


`$<1:...>` or `$<0:...>`: The result of the inner expression is used to form a new expression.

- If the config was `Debug`, the expression becomes `$<1:IS_DEBUG_BUILD>`. The `$<1:...>` expression simply evaluates to its content, so the result is the string `"IS_DEBUG_BUILD"`.
- If the config was not `Debug` (e.g., `Release`), the expression becomes `$<0:IS_DEBUG_BUILD>`. The `$<0:...>` expression evaluates to an empty string.

So, for a `Debug` build, the command effectively becomes:


```javascript
target_compile_definitions(GreeterApp PRIVATE IS_DEBUG_BUILD)
```


 For any other build type, the expression resolves to the following, which effectively does nothing:


```javascript
target_compile_definitions(GreeterApp PRIVATE "")
```


### The `IF` Expression


This is a more general conditional expression, like a ternary operator. It evaluates a condition (which is `1` for true, `0` for false) and resolves to one of two strings.


A classic use case is selecting the correct C++ runtime library when using MSVC. Debug builds should link against the debug runtime (`/MDd`), while release builds should use the release runtime (`/MD`).


```c++
#<cmake>
# This only applies the flags if the compiler is MSVC
if(MSVC)
  target_compile_options(GreeterApp PRIVATE
    $<$<IF:$<CONFIG:Debug>,/MDd,/MD>>
  )
endif()
```


Here's how it works at build time:

1. `$<CONFIG:Debug>` evaluates to `1` or `0`.
2. The expression becomes either `$<IF:1,/MDd,/MD>` or `$<IF:0,/MDd,/MD>`.
3. `$<IF:1,...>` evaluates to the first string (`/MDd`).
4. `$<IF:0,...>` evaluates to the second string (`/MD`).

### The `AND`, `OR`, and `NOT` Expressions


Just like `if()` statements, generator expressions can be combined using logical operators to create more complex conditions. These expressions take a semicolon-separated list of other boolean expressions (ones that evaluate to `0` or `1`) as their arguments.

- `$<AND:expressions...>`: Evaluates to `1` only if _**all**_ of the contained expressions evaluate to `1`.
- `$<OR:expressions...>`: Evaluates to `1` if _**any**_ of the contained expressions evaluate to `1`.
- `$<NOT:expression>`: Evaluates to `1` if the single contained expression is `0`, and `0` if it is `1`.

Let's use this to set a specific preprocessor definition only when we are building in `Debug` mode **and** on the Windows platform.


Here, `$<AND:...>` will only resolve to `1` if **both** `$<CONFIG:Debug>` and `$<PLATFORM_ID:Windows>` are true.  We can add additional spacing to complex generator expressions to make them easier to follow:


```c++
//<cmake>
target_compile_definitions(MyApp PRIVATE
  $<
    $<AND:
      $<CONFIG:Debug>,
      $<PLATFORM_ID:Windows>
    >:
    IS_WINDOWS_DEBUG_BUILD
  >
)
```


We'll introduce alternative ways to simplify this in the next section.


### The `TARGET_FILE` Expression


Sometimes, you need to know the full path to the output file of a target. For example, you might want to run a custom command that processes an executable after it has been built.


Hardcoding that location by typing something like `/build/app/GreeterApp.exe` isn't recommended, as it's not portable.  The output name and location can change based on the generator, configuration, and platform.


The `$<TARGET_FILE:SomeTarget>` expression solves this. It resolves, at build time, to the full path of the specified target's main file.  In the following example, we use this to log out the full path of our executable, after it is built.


The `add_custom_command()` syntax in this example is a new concept that we will cover in the next lesson. For now, the key point is that `$<TARGET_FILE:SomeTarget>` expression provides a reliable way to get a target's output path:


<CodeViewer>


<CodeFile path="CMakeLists.txt">


```c++
#<cmake>
cmake_minimum_required(VERSION 3.16)
project(Greeter)

add_subdirectory(app)
```


</CodeFile>


<CodeFile path="app/CMakeLists.txt" defaultOpen>


```c++
#<cmake>
cmake_minimum_required(VERSION 3.16)

add_executable(GreeterApp src/main.cpp)

add_custom_command(TARGET GreeterApp
  POST_BUILD COMMAND ${CMAKE_COMMAND} -E echo
  "Built binary located at $<TARGET_FILE:GreeterApp>"#<h>
)
```


</CodeFile>


<CodeFile path="app/src/main.cpp">


```c++
#include <iostream>

int main() {
  std::cout << "Hello World";
  return 0;
}
```


</CodeFile>


</CodeViewer>


If we run a clean build, we should see our generator expression outputting the location of our executable:


```c++
//<ti>
cmake --build .
```


```c++
//<o>
...
Linking CXX executable app\GreeterApp.exe
Built binary located at D:/Projects/cmake/build/app/GreeterApp.exe
```


### The `PLATFORM_ID` Expression


This expression works just like `$<CONFIG:...>`, but it checks the target platform identifier (e.g., `Windows`, `Linux`, `Darwin` for macOS). It provides a build-time alternative to the configure-time `if(WIN32)` check.


In the following example, we define the `IS_WINDOWS_BUILD` preprocessor macro, but only we're building for Windows.


```c++
#<cmake>
# Add a definition only when building for Windows
target_compile_definitions(MyTarget PRIVATE
  $<$<PLATFORM_ID:Windows>:IS_WINDOWS_BUILD>
)
```


This is less common than a check such as `if(WIN32)`, but can be useful if the platform decision needs to be deferred to build time, especially in complex cross-compilation scenarios.


<Pw>


## Best Practices and Writing Clean Expressions


Generator expressions are powerful, but complex, nested expressions can quickly become unreadable.  The following expression sets the `-some-flag` compiler option if the build type is Debug **and** the platform is Windows, and `-other-flag` otherwise:


```c++
#<cmake>
# Hard to read and maintain
target_compile_options(MyApp PRIVATE
  "$<IF:$<AND:$<CONFIG:Debug>,$<PLATFORM_ID:Windows>>,-some-flag,-other-flag>")
```


Here are some good practices for keeping your scripts clean.


### Use Variables for Readability


The single most effective way to clean up a complex generator expression is to store it in a variable. This gives the logic a descriptive name and separates the complex condition from the command that uses it.


Let's refactor our "bad" example using this technique:


```c++
#<cmake>
# Good: Store the condition in a variable
set(
  IS_WINDOWS_DEBUG
  "$<AND:$<CONFIG:Debug>,$<PLATFORM_ID:Windows>>"
)

# Use the variable in a clean $<IF:...> expression
target_compile_options(MyApp PRIVATE
  "$<IF:${IS_WINDOWS_DEBUG},-some-flag,-other-flag>"
)
```


The result is the same, but the `target_compile_options()` call is now much easier to understand at a glance. We've given the complex condition a name, `IS_WINDOWS_DEBUG`, which documents its purpose.


### Use Whitespace for Formatting


Don't be afraid to use whitespace. CMake is flexible about newlines inside a command's parentheses. You can format long generator expressions across multiple lines to improve clarity, just like you would with complex C++ code.


```c++
#<cmake>
# Good: Use formatting for clarity
target_compile_options(MyApp PRIVATE
  "$<IF:
    $<AND:
      $<CONFIG:Debug>,
      $<PLATFORM_ID:Windows>
    >,
    -some-flag,
    -other-flag
  >"
)
```


This formatted version is much easier to parse visually than a single, long line. You can clearly see the structure of the nested expressions.


### Encapsulate Logic in `INTERFACE` Targets


Remember, for logic that is reused across multiple targets, it's often best to encapsulate it within an `INTERFACE` target. This is the ultimate expression of the "Don't Repeat Yourself" (DRY) principle in CMake.


We walked through an example of creating and using `INTERFACE` targets <InlineLessonPromo url="cmake/cmake-property-groups-aliases-and-imports" title="Using `INTERFACE`, `ALIAS`, and `IMPORTED` Libraries" description="Learn to use abstract target types like `INTERFACE`, `ALIAS`, and `IMPORTED` to model complex project needs, organize build properties, and integrate pre-compiled binaries.">earlier in the course</InlineLessonPromo>.  The key point is that, Instead of having multiple targets repeat the logic to decide which properties should be applied, we can do that in one place to create a single "property bag" target:


```c++
#<cmake>
# Create an INTERFACE target to hold the complex logic
add_library(BuildPolicies INTERFACE)

# Attach the generator expression to the interface target
set(
  IS_WINDOWS_DEBUG
  "$<AND:$<CONFIG:Debug>,$<PLATFORM_ID:Windows>>"
)

target_compile_options(BuildPolicies PRIVATE
  "$<IF:${IS_WINDOWS_DEBUG},-some-flag,-other-flag>"
)
```


The complex logic is defined in exactly one place. Consumers don't need to know the details; they just link to the `BuildPolicies` abstraction.


```c++
#<cmake>
target_link_libraries(App1 PRIVATE BuildPolicies)
target_link_libraries(App2 PRIVATE BuildPolicies)
```


## Summary


Generator expressions provide a mechanism for embedding deferred logic directly into the generated build files.

- **Configure-Time vs. Build-Time:** `if()` statements are evaluated when `cmake` runs. Generator expressions (`$<...>`) are evaluated when you build (e.g., when `make` or `MSBuild` runs).
- **Portability is Key:** Using generator expressions lets us write a single `CMakeLists.txt` that works correctly for both single-configuration (Makefiles) and multi-configuration (Visual Studio) generators.
- **Best Practice:** Prefer generator expressions over `if()` statements for any logic that depends on the build configuration. To keep them clean, use variables to name complex conditions, format them with whitespace, and encapsulate reused logic in `INTERFACE` targets.

Some of the most useful expressions include:

- `CONFIG` is for configuration-specific logic (`Debug`, `Release`, etc.).
- `IF` provides a general-purpose conditional.
- `AND`, `OR`, and `NOT` let us build more complex checks
- `TARGET_FILE` gives you the path to a target's output file.
- `PLATFORM_ID` give us an alternative to platform checks like `if(WIN32)` that can be evaluated at build time

<Preview>


### Custom Build Steps


We've seen how `$<TARGET_FILE:...>` can be used to get the path to a built file, but we haven't fully explored what we can do with that information. What if you need to run a code generator, copy a file after it's built, or package your assets?


In the next lessons, we will introduce `add_custom_command()` and `add_custom_target()`, the CMake commands that let you add arbitrary scripts and tools to your build process.


</Preview>


</Pw>

Custom Commands for Build Steps

cmake-generator-expressions(7) Manual

`CMAKE_BUILD_TYPE` Variable

It's Time To Do CMake Right - Generator Expressions

More Modern CMake - Deniz Bahadir

Effective Modern CMake

Daniel Pfeifer: Effective CMake

Master modern CMake by learning generator expressions. This guide explains why 'if(CMAKE_BUILD_TYPE)' fails for multi-config generators and how to use $CONFIG:... and $IF:... for portable, build-time conditional logic.

CMake Generator Expressions: Modern Conditional Logic for C++ Builds

Best Practices and Writing Clean Expressions

Generator Expressions and Conditional Logic

Understanding the Basics

Practical Applications

Common Mistakes

Next Steps

Additional Resources

Assessment Preparation

Managing C++ Projects Using CMake