glfw/docs/compile.md

Compiling GLFW

@tableofcontents

This is about compiling the GLFW library itself. For information on how to build applications that use GLFW, see @ref build_guide.

GLFW uses some C99 features and does not support Visual Studio 2012 and earlier.

Using CMake

GLFW behaves like most other libraries that use CMake so this guide mostly describes the standard configure, generate and compile sequence. If you are already familiar with this from other projects, you may want to focus on the @ref compile_deps and @ref compile_options sections for GLFW-specific information.

GLFW uses CMake to generate project files or makefiles for your chosen development environment. To compile GLFW, first generate these files with CMake and then use them to compile the GLFW library.

If you are on Windows and macOS you can download CMake from their site.

If you are on a Unix-like system such as Linux, FreeBSD or Cygwin or have a package system like Fink, MacPorts or Homebrew, you can install its CMake package.

CMake is a complex tool and this guide will only show a few of the possible ways to set up and compile GLFW. The CMake project has their own much more detailed CMake user guide that includes everything in this guide not specific to GLFW. It may be a useful companion to this one.

Installing dependencies

The C/C++ development environments in Visual Studio, Xcode and MinGW come with all necessary dependencies for compiling GLFW, but on Unix-like systems like Linux and FreeBSD you will need a few extra packages.

Dependencies for Wayland and X11

By default, both the Wayland and X11 backends are enabled on Linux and other Unix-like systems (except macOS). To disable one or both of these, set the @ref GLFW_BUILD_WAYLAND or @ref GLFW_BUILD_X11 CMake options in the next step when generating build files.

To compile GLFW for both Wayland and X11, you need to have the X11, Wayland and xkbcommon development packages installed. On some systems a few other packages are also required. None of the development packages above are needed to build or run programs that use an already compiled GLFW library.

On Debian and derivatives like Ubuntu and Linux Mint you will need the libwayland-dev and libxkbcommon-dev packages to compile for Wayland and the xorg-dev meta-package to compile for X11. These will pull in all other dependencies.

@code{.sh} sudo apt install libwayland-dev libxkbcommon-dev xorg-dev @endcode

On Fedora and derivatives like Red Hat you will need the wayland-devel and libxkbcommon-devel packages to compile for Wayland and the libXcursor-devel, libXi-devel, libXinerama-devel and libXrandr-devel packages to compile for X11. These will pull in all other dependencies.

@code{.sh} sudo dnf install wayland-devel libxkbcommon-devel libXcursor-devel libXi-devel libXinerama-devel libXrandr-devel @endcode

On FreeBSD you will need the wayland, libxkbcommon and evdev-proto packages to compile for Wayland. The X11 headers are installed along the end-user X11 packages, so if you have an X server running you should have the headers as well. If not, install the xorgproto package to compile for X11.

@code{.sh} pkg install wayland libxkbcommon evdev-proto xorgproto @endcode

On Cygwin Wayland is not supported but you will need the libXcursor-devel, libXi-devel, libXinerama-devel, libXrandr-devel and libXrender-devel packages to compile for X11. These can be found in the Libs section of the GUI installer and will pull in all other dependencies.

Once you have the required dependencies, move on to @ref compile_generate.

Generating build files with CMake

Once you have all necessary dependencies it is time to generate the project files or makefiles for your development environment. CMake needs two paths for this:

• the path to the root directory of the GLFW source tree (not its src subdirectory)
• the path to the directory where the generated build files and compiled binaries will be placed

If these are the same, it is called an in-tree build, otherwise it is called an out-of-tree build.

Out-of-tree builds are recommended as they avoid cluttering up the source tree. They also allow you to have several build directories for different configurations all using the same source tree.

A common pattern when building a single configuration is to have a build directory named build in the root of the source tree.

Generating with the CMake GUI

Start the CMake GUI and set the paths to the source and build directories described above. Then press Configure and Generate.

If you wish change any CMake variables in the list, press Configure and then Generate to have the new values take effect. The variable list will be populated after the first configure step.

By default, GLFW will use Wayland and X11 on Linux and other Unix-like systems other than macOS. To disable support for one or both of these, set the @ref GLFW_BUILD_WAYLAND and/or @ref GLFW_BUILD_X11 option in the GLFW section of the variable list, then apply the new value as described above.

Once you have generated the project files or makefiles for your chosen development environment, move on to @ref compile_compile.

Generating with command-line CMake

To make a build directory, pass the source and build directories to the cmake command. These can be relative or absolute paths. The build directory is created if it doesn't already exist.

@code{.sh} cmake -S path/to/glfw -B path/to/build @endcode

It is common to name the build directory build and place it in the root of the source tree when only planning to build a single configuration.

@code{.sh} cd path/to/glfw cmake -S . -B build @endcode

Without other flags these will generate Visual Studio project files on Windows and makefiles on other platforms. You can choose other targets using the -G flag.

@code{.sh} cmake -S path/to/glfw -B path/to/build -G Xcode @endcode

By default, GLFW will use Wayland and X11 on Linux and other Unix-like systems other than macOS. To disable support for one or both of these, set the @ref GLFW_BUILD_WAYLAND and/or @ref GLFW_BUILD_X11 CMake option.

@code{.sh} cmake -S path/to/glfw -B path/to/build -D GLFW_BUILD_X11=0 @endcode

Once you have generated the project files or makefiles for your chosen development environment, move on to @ref compile_compile.

Compiling the library

You should now have all required dependencies and the project files or makefiles necessary to compile GLFW. Go ahead and compile the actual GLFW library with these files as you would with any other project.

With Visual Studio open GLFW.sln and use the Build menu. With Xcode open GLFW.xcodeproj and use the Project menu.

With Linux, macOS and other forms of Unix, run make.

@code{.sh} cd path/to/build make @endcode

With MinGW, it is mingw32-make.

@code{.sh} cd path/to/build mingw32-make @endcode

Any CMake build directory can also be built with the cmake command and the --build flag.

@code{.sh} cmake --build path/to/build @endcode

This will run the platform specific build tool the directory was generated for.

Once the GLFW library is compiled you are ready to build your application, linking it to the GLFW library. See @ref build_guide for more information.

CMake options

The CMake files for GLFW provide a number of options, although not all are available on all supported platforms. Some of these are de facto standards among projects using CMake and so have no GLFW_ prefix.

If you are using the GUI version of CMake, these are listed and can be changed from there. If you are using the command-line version of CMake you can use the ccmake ncurses GUI to set options. Some package systems like Ubuntu and other distributions based on Debian GNU/Linux have this tool in a separate cmake-curses-gui package.

Finally, if you don't want to use any GUI, you can set options from the cmake command-line with the -D flag.

@code{.sh} cmake -S path/to/glfw -B path/to/build -D BUILD_SHARED_LIBS=ON @endcode

Shared CMake options

@anchor BUILD_SHARED_LIBS BUILD_SHARED_LIBS determines whether GLFW is built as a static library or as a DLL / shared library / dynamic library. This is disabled by default, producing a static GLFW library. This variable has no GLFW_ prefix because it is defined by CMake. If you want to change the library only for GLFW when it is part of a larger project, see @ref GLFW_LIBRARY_TYPE.

@anchor GLFW_LIBRARY_TYPE GLFW_LIBRARY_TYPE allows you to override @ref BUILD_SHARED_LIBS only for GLFW, without affecting other libraries in a larger project. When set, the value of this option must be a valid CMake library type. Set it to STATIC to build GLFW as a static library, SHARED to build it as a shared library / dynamic library / DLL, or OBJECT to make GLFW a CMake object library.

@anchor GLFW_BUILD_EXAMPLES GLFW_BUILD_EXAMPLES determines whether the GLFW examples are built along with the library. This is enabled by default unless GLFW is being built as a subproject of a larger CMake project.

@anchor GLFW_BUILD_TESTS GLFW_BUILD_TESTS determines whether the GLFW test programs are built along with the library. This is enabled by default unless GLFW is being built as a subproject of a larger CMake project.

@anchor GLFW_BUILD_DOCS GLFW_BUILD_DOCS determines whether the GLFW documentation is built along with the library. This is enabled by default if Doxygen is found by CMake during configuration.

Win32 specific CMake options

@anchor GLFW_BUILD_WIN32 GLFW_BUILD_WIN32 determines whether to include support for Win32 when compiling the library. This option is only available when compiling for Windows. This is enabled by default.

@anchor USE_MSVC_RUNTIME_LIBRARY_DLL USE_MSVC_RUNTIME_LIBRARY_DLL determines whether to use the DLL version or the static library version of the Visual C++ runtime library. When enabled, the DLL version of the Visual C++ library is used. This is enabled by default.

On CMake 3.15 and later you can set the standard CMake CMAKE_MSVC_RUNTIME_LIBRARY variable instead of this GLFW-specific option.

@anchor GLFW_USE_HYBRID_HPG GLFW_USE_HYBRID_HPG determines whether to export the NvOptimusEnablement and AmdPowerXpressRequestHighPerformance symbols, which force the use of the high-performance GPU on Nvidia Optimus and AMD PowerXpress systems. These symbols need to be exported by the EXE to be detected by the driver, so the override will not work if GLFW is built as a DLL. This is disabled by default, letting the operating system and driver decide.

macOS specific CMake options

@anchor GLFW_BUILD_COCOA GLFW_BUILD_COCOA determines whether to include support for Cocoa when compiling the library. This option is only available when compiling for macOS. This is enabled by default.

Unix-like system specific CMake options

@anchor GLFW_BUILD_WAYLAND GLFW_BUILD_WAYLAND determines whether to include support for Wayland when compiling the library. This option is only available when compiling for Linux and other Unix-like systems other than macOS. This is enabled by default.

@anchor GLFW_BUILD_X11 GLFW_BUILD_X11 determines whether to include support for X11 when compiling the library. This option is only available when compiling for Linux and other Unix-like systems other than macOS. This is enabled by default.

Cross-compilation with CMake and MinGW

Both Cygwin and many Linux distributions have MinGW or MinGW-w64 packages. For example, Cygwin has the mingw64-i686-gcc and mingw64-x86_64-gcc packages for 32- and 64-bit version of MinGW-w64, while Debian GNU/Linux and derivatives like Ubuntu have the mingw-w64 package for both.

GLFW has CMake toolchain files in the CMake subdirectory that set up cross-compilation of Windows binaries. To use these files you set the CMAKE_TOOLCHAIN_FILE CMake variable with the -D flag add an option when configuring and generating the build files.

@code{.sh} cmake -S path/to/glfw -B path/to/build -D CMAKE_TOOLCHAIN_FILE=path/to/file @endcode

The exact toolchain file to use depends on the prefix used by the MinGW or MinGW-w64 binaries on your system. You can usually see this in the /usr directory. For example, both the Ubuntu and Cygwin MinGW-w64 packages have /usr/x86_64-w64-mingw32 for the 64-bit compilers, so the correct invocation would be:

@code{.sh} cmake -S path/to/glfw -B path/to/build -D CMAKE_TOOLCHAIN_FILE=CMake/x86_64-w64-mingw32.cmake @endcode

The path to the toolchain file is relative to the path to the GLFW source tree passed to the -S flag, not to the current directory.

For more details see the CMake toolchain guide.

Compiling GLFW manually

If you wish to compile GLFW without its CMake build environment then you will have to do at least some platform-detection yourself. There are preprocessor macros for enabling support for the platforms (window systems) available. There are also optional, platform-specific macros for various features.

When building, GLFW will expect the necessary configuration macros to be defined on the command-line. The GLFW CMake files set these as private compile definitions on the GLFW target but if you compile the GLFW sources manually you will need to define them yourself.

The window system is used to create windows, handle input, monitors, gamma ramps and clipboard. The options are:

• @b _GLFW_COCOA to use the Cocoa frameworks
• @b _GLFW_WIN32 to use the Win32 API
• @b _GLFW_X11 to use the X Window System
• @b _GLFW_WAYLAND to use the Wayland API (incomplete)

The @b _GLFW_WAYLAND and @b _GLFW_X11 macros may be combined and produces a library that attempts to detect the appropriate platform at initialization.

If you are building GLFW as a shared library / dynamic library / DLL then you must also define @b _GLFW_BUILD_DLL. Otherwise, you must not define it.

If you are using a custom name for the Vulkan, EGL, GLX, OSMesa, OpenGL, GLESv1 or GLESv2 library, you can override the default names by defining those you need of @b _GLFW_VULKAN_LIBRARY, @b _GLFW_EGL_LIBRARY, @b _GLFW_GLX_LIBRARY, @b _GLFW_OSMESA_LIBRARY, @b _GLFW_OPENGL_LIBRARY, @b _GLFW_GLESV1_LIBRARY and @b _GLFW_GLESV2_LIBRARY. Otherwise, GLFW will use the built-in default names.

@note None of the @ref build_macros may be defined during the compilation of GLFW. If you define any of these in your build files, make sure they are not applied to the GLFW sources.