- Cmake linux environment variables
- Set Normal VariableВ¶
- Set Cache EntryВ¶
- Set Environment VariableВ¶
- User Interaction GuideВ¶
- Command Line cmake toolВ¶
- cmake-gui toolВ¶
- Generating a BuildsystemВ¶
- Command line environmentВ¶
- Command line -G optionВ¶
- Choosing a generator in cmake-guiВ¶
- Setting Build VariablesВ¶
- Setting variables on the command lineВ¶
- Setting variables with cmake-guiВ¶
- The CMake CacheВ¶
- Using presets on the command-lineВ¶
- Using presets in cmake-guiВ¶
- Invoking the BuildsystemВ¶
- Selecting a TargetВ¶
- Specifying a Build ProgramВ¶
Cmake linux environment variables
Set a normal, cache, or environment variable to a given value. See the cmake-language(7) variables documentation for the scopes and interaction of normal variables and cache entries.
Signatures of this command that specify a . placeholder expect zero or more arguments. Multiple arguments will be joined as a semicolon-separated list to form the actual variable value to be set. Zero arguments will cause normal variables to be unset. See the unset() command to unset variables explicitly.
Set Normal VariableВ¶
Sets the given in the current function or directory scope.
If the PARENT_SCOPE option is given the variable will be set in the scope above the current scope. Each new directory or function() command creates a new scope. A scope can also be created with the block() command. This command will set the value of a variable into the parent directory, calling function or encompassing scope (whichever is applicable to the case at hand). The previous state of the variable’s value stays the same in the current scope (e.g., if it was undefined before, it is still undefined and if it had a value, it is still that value).
The block(PROPAGATE) and return(PROPAGATE) commands can be used as an alternate method to the set(PARENT_SCOPE) and unset(PARENT_SCOPE) commands to update the parent scope.
Set Cache EntryВ¶
Sets the given cache (cache entry). Since cache entries are meant to provide user-settable values this does not overwrite existing cache entries by default. Use the FORCE option to overwrite existing entries.
The must be specified as one of:
Boolean ON/OFF value. cmake-gui(1) offers a checkbox.
Path to a file on disk. cmake-gui(1) offers a file dialog.
Path to a directory on disk. cmake-gui(1) offers a file dialog.
A line of text. cmake-gui(1) offers a text field or a drop-down selection if the STRINGS cache entry property is set.
A line of text. cmake-gui(1) does not show internal entries. They may be used to store variables persistently across runs. Use of this type implies FORCE .
The must be specified as a line of text providing a quick summary of the option for presentation to cmake-gui(1) users.
If the cache entry does not exist prior to the call or the FORCE option is given then the cache entry will be set to the given value.
The content of the cache variable will not be directly accessible if a normal variable of the same name already exists (see rules of variable evaluation ). If policy CMP0126 is set to OLD , any normal variable binding in the current scope will be removed.
It is possible for the cache entry to exist prior to the call but have no type set if it was created on the cmake(1) command line by a user through the -D = option without specifying a type. In this case the set command will add the type. Furthermore, if the is PATH or FILEPATH and the provided on the command line is a relative path, then the set command will treat the path as relative to the current working directory and convert it to an absolute path.
Set Environment VariableВ¶
Sets an Environment Variable to the given value. Subsequent calls of $ENV < >will return this new value.
This command affects only the current CMake process, not the process from which CMake was called, nor the system environment at large, nor the environment of subsequent build or test processes.
If no argument is given after ENV < >or if is an empty string, then this command will clear any existing value of the environment variable.
Arguments after are ignored. If extra arguments are found, then an author warning is issued.
User Interaction GuideВ¶
Where a software package supplies a CMake-based buildsystem with the source of their software, the consumer of the software is required to run a CMake user interaction tool in order to build it.
Well-behaved CMake-based buildsystems do not create any output in the source directory, so typically, the user performs an out-of-source build and performs the build there. First, CMake must be instructed to generate a suitable buildsystem, then the user invokes a build tool to process that generated buildsystem. The generated buildsystem is specific to the machine used to generate it and is not redistributable. Each consumer of a provided source software package is required to use CMake to generate a buildsystem specific to their system.
Generated buildsystems should generally be treated as read-only. The CMake files as a primary artifact should completely specify the buildsystem and there should be no reason to populate properties manually in an IDE for example after generating the buildsystem. CMake will periodically rewrite the generated buildsystem, so modifications by users will be overwritten.
The features and user interfaces described in this manual are available for all CMake-based build systems by virtue of providing CMake files.
The CMake tooling may report errors to the user when processing provided CMake files, such as reporting that the compiler is not supported, or the compiler does not support a required compile option, or a dependency can not be found. These errors must be resolved by the user by choosing a different compiler, installing dependencies , or instructing CMake where to find them, etc.
Command Line cmake toolВ¶
A simple but typical use of cmake(1) with a fresh copy of software source code is to create a build directory and invoke cmake there:
It is recommended to build in a separate directory to the source because that keeps the source directory pristine, allows for building a single source with multiple toolchains, and allows easy clearing of build artifacts by simply deleting the build directory.
The CMake tooling may report warnings which are intended for the provider of the software, not intended for the consumer of the software. Such warnings end with «This warning is for project developers». Users may disable such warnings by passing the -Wno-dev flag to cmake(1) .
Users more accustomed to GUI interfaces may use the cmake-gui(1) tool to invoke CMake and generate a buildsystem.
The source and binary directories must first be populated. It is always advised to use different directories for the source and the build.
Generating a BuildsystemВ¶
There are several user interface tools which may be used to generate a buildsystem from CMake files. The ccmake(1) and cmake-gui(1) tools guide the user through setting the various necessary options. The cmake(1) tool can be invoked to specify options on the command line. This manual describes options which may be set using any of the user interface tools, though the mode of setting an option is different for each tool.
Command line environmentВ¶
When invoking cmake(1) with a command line buildsystem such as Makefiles or Ninja , it is necessary to use the correct build environment to ensure that build tools are available. CMake must be able to find the appropriate build tool , compiler, linker and other tools as needed.
On Linux systems, the appropriate tools are often provided in system-wide locations and may be readily installed through the system package manager. Other toolchains provided by the user or installed in non-default locations can also be used.
When cross-compiling, some platforms may require environment variables to be set or may provide scripts to set the environment.
Visual Studio ships multiple command prompts and vcvarsall.bat scripts for setting up the correct environments for command line buildsystems. While not strictly necessary to use a corresponding command line environment when using a Visual Studio generator, doing so has no disadvantages.
When using Xcode, there can be more than one Xcode version installed. Which one to use can be selected in a number of different ways, but the most common methods are:
Setting the default version in the preferences of the Xcode IDE.
Setting the default version via the xcode-select command line tool.
Overriding the default version by setting the DEVELOPER_DIR environment variable when running CMake and the build tool.
For convenience, cmake-gui(1) provides an environment variable editor.
Command line -G optionВ¶
CMake chooses a generator by default based on the platform. Usually, the default generator is sufficient to allow the user to proceed to build the software.
The user may override the default generator with the -G option:
The output of cmake —help includes a list of generators available for the user to choose from. Note that generator names are case sensitive.
On Unix-like systems (including Mac OS X), the Unix Makefiles generator is used by default. A variant of that generator can also be used on Windows in various environments, such as the NMake Makefiles and MinGW Makefiles generator. These generators generate a Makefile variant which can be executed with make , gmake , nmake or similar tools. See the individual generator documentation for more information on targeted environments and tools.
The Ninja generator is available on all major platforms. ninja is a build tool similar in use-cases to make , but with a focus on performance and efficiency.
On Windows, cmake(1) can be used to generate solutions for the Visual Studio IDE. Visual Studio versions may be specified by the product name of the IDE, which includes a four-digit year. Aliases are provided for other means by which Visual Studio versions are sometimes referred to, such as two digits which correspond to the product version of the VisualC++ compiler, or a combination of the two:
Visual Studio generators can target different architectures. One can specify the target architecture using the -A option:
On Apple, the Xcode generator may be used to generate project files for the Xcode IDE.
Some IDEs such as KDevelop4, QtCreator and CLion have native support for CMake-based buildsystems. Those IDEs provide user interface for selecting an underlying generator to use, typically a choice between a Makefile or a Ninja based generator.
Note that it is not possible to change the generator with -G after the first invocation of CMake. To change the generator, the build directory must be deleted and the build must be started from scratch.
When generating Visual Studio project and solutions files several other options are available to use when initially running cmake(1) .
The Visual Studio toolset can be specified with the cmake -T option:
Whereas the -A option specifies the _target_ architecture, the -T option can be used to specify details of the toolchain used. For example, -Thost=x64 can be given to select the 64-bit version of the host tools. The following demonstrates how to use 64-bit tools and also build for a 64-bit target architecture:
Choosing a generator in cmake-guiВ¶
The «Configure» button triggers a new dialog to select the CMake generator to use.
All generators available on the command line are also available in cmake-gui(1) .
When choosing a Visual Studio generator, further options are available to set an architecture to generate for.
Setting Build VariablesВ¶
Software projects often require variables to be set on the command line when invoking CMake. Some of the most commonly used CMake variables are listed in the table below:
Path to search for additional CMake modules
Build configuration, such as Debug or Release , determining debug/optimization flags. This is only relevant for single-configuration buildsystems such as Makefile and Ninja . Multi-configuration buildsystems such as those for Visual Studio and Xcode ignore this setting.
Location to install the software to with the install build target
File containing cross-compiling data such as toolchains and sysroots .
Whether to build shared instead of static libraries for add_library() commands used without a type
Generate a compile_commands.json file for use with clang-based tools
Other project-specific variables may be available to control builds, such as enabling or disabling components of the project.
There is no convention provided by CMake for how such variables are named between different provided buildsystems, except that variables with the prefix CMAKE_ usually refer to options provided by CMake itself and should not be used in third-party options, which should use their own prefix instead. The cmake-gui(1) tool can display options in groups defined by their prefix, so it makes sense for third parties to ensure that they use a self-consistent prefix.
Setting variables on the command lineВ¶
CMake variables can be set on the command line either when creating the initial build:
or later on a subsequent invocation of cmake(1) :
The -U flag may be used to unset variables on the cmake(1) command line:
A CMake buildsystem which was initially created on the command line can be modified using the cmake-gui(1) and vice-versa.
The cmake(1) tool allows specifying a file to use to populate the initial cache using the -C option. This can be useful to simplify commands and scripts which repeatedly require the same cache entries.
Setting variables with cmake-guiВ¶
Variables may be set in the cmake-gui using the «Add Entry» button. This triggers a new dialog to set the value of the variable.
The main view of the cmake-gui(1) user interface can be used to edit existing variables.
The CMake CacheВ¶
When CMake is executed, it needs to find the locations of compilers, tools and dependencies. It also needs to be able to consistently re-generate a buildsystem to use the same compile/link flags and paths to dependencies. Such parameters are also required to be configurable by the user because they are paths and options specific to the users system.
When it is first executed, CMake generates a CMakeCache.txt file in the build directory containing key-value pairs for such artifacts. The cache file can be viewed or edited by the user by running the cmake-gui(1) or ccmake(1) tool. The tools provide an interactive interface for re-configuring the provided software and re-generating the buildsystem, as is needed after editing cached values. Each cache entry may have an associated short help text which is displayed in the user interface tools.
The cache entries may also have a type to signify how it should be presented in the user interface. For example, a cache entry of type BOOL can be edited by a checkbox in a user interface, a STRING can be edited in a text field, and a FILEPATH while similar to a STRING should also provide a way to locate filesystem paths using a file dialog. An entry of type STRING may provide a restricted list of allowed values which are then provided in a drop-down menu in the cmake-gui(1) user interface (see the STRINGS cache property).
The CMake files shipped with a software package may also define boolean toggle options using the option() command. The command creates a cache entry which has a help text and a default value. Such cache entries are typically specific to the provided software and affect the configuration of the build, such as whether tests and examples are built, whether to build with exceptions enabled etc.
CMake understands a file, CMakePresets.json , and its user-specific counterpart, CMakeUserPresets.json , for saving presets for commonly-used configure settings. These presets can set the build directory, generator, cache variables, environment variables, and other command-line options. All of these options can be overridden by the user. The full details of the CMakePresets.json format are listed in the cmake-presets(7) manual.
Using presets on the command-lineВ¶
When using the cmake(1) command line tool, a preset can be invoked by using the —preset option. If —preset is specified, the generator and build directory are not required, but can be specified to override them. For example, if you have the following CMakePresets.json file:
and you run the following:
This will generate a build directory in /path/to/source/build/ninja-release with the Ninja generator, and with CMAKE_BUILD_TYPE set to Release .
If you want to see the list of available presets, you can run:
This will list the presets available in /path/to/source/CMakePresets.json and /path/to/source/CMakeUsersPresets.json without generating a build tree.
Using presets in cmake-guiВ¶
If a project has presets available, either through CMakePresets.json or CMakeUserPresets.json , the list of presets will appear in a drop-down menu in cmake-gui(1) between the source directory and the binary directory. Choosing a preset sets the binary directory, generator, environment variables, and cache variables, but all of these options can be overridden after a preset is selected.
Invoking the BuildsystemВ¶
After generating the buildsystem, the software can be built by invoking the particular build tool. In the case of the IDE generators, this can involve loading the generated project file into the IDE to invoke the build.
CMake is aware of the specific build tool needed to invoke a build so in general, to build a buildsystem or project from the command line after generating, the following command may be invoked in the build directory:
The —build flag enables a particular mode of operation for the cmake(1) tool. It invokes the CMAKE_MAKE_PROGRAM command associated with the generator , or the build tool configured by the user.
The —build mode also accepts the parameter —target to specify a particular target to build, for example a particular library, executable or custom target, or a particular special target like install :
The —build mode also accepts a —config parameter in the case of multi-config generators to specify which particular configuration to build:
The —config option has no effect if the generator generates a buildsystem specific to a configuration which is chosen when invoking cmake with the CMAKE_BUILD_TYPE variable.
Some buildsystems omit details of command lines invoked during the build. The —verbose flag can be used to cause those command lines to be shown:
The —build mode can also pass particular command line options to the underlying build tool by listing them after — . This can be useful to specify options to the build tool, such as to continue the build after a failed job, where CMake does not provide a high-level user interface.
For all generators, it is possible to run the underlying build tool after invoking CMake. For example, make may be executed after generating with the Unix Makefiles generator to invoke the build, or ninja after generating with the Ninja generator etc. The IDE buildsystems usually provide command line tooling for building a project which can also be invoked.
Selecting a TargetВ¶
Each executable and library described in the CMake files is a build target, and the buildsystem may describe custom targets, either for internal use, or for user consumption, for example to create documentation.
CMake provides some built-in targets for all buildsystems providing CMake files.
The default target used by Makefile and Ninja generators. Builds all targets in the buildsystem, except those which are excluded by their EXCLUDE_FROM_ALL target property or EXCLUDE_FROM_ALL directory property. The name ALL_BUILD is used for this purpose for the Xcode and Visual Studio generators.
Lists the targets available for build. This target is available when using the Unix Makefiles or Ninja generator, and the exact output is tool-specific.
Delete built object files and other output files. The Makefile based generators create a clean target per directory, so that an individual directory can be cleaned. The Ninja tool provides its own granular -t clean system.
Runs tests. This target is only automatically available if the CMake files provide CTest-based tests. See also Running Tests.
Installs the software. This target is only automatically available if the software defines install rules with the install() command. See also Software Installation.
Creates a binary package. This target is only automatically available if the CMake files provide CPack-based packages.
Creates a source package. This target is only automatically available if the CMake files provide CPack-based packages.
For Makefile based systems, /fast variants of binary build targets are provided. The /fast variants are used to build the specified target without regard for its dependencies. The dependencies are not checked and are not rebuilt if out of date. The Ninja generator is sufficiently fast at dependency checking that such targets are not provided for that generator.
Makefile based systems also provide build-targets to preprocess, assemble and compile individual files in a particular directory.
The file extension is built into the name of the target because another file with the same name but a different extension may exist. However, build-targets without the file extension are also provided.
In buildsystems which contain foo.c and foo.cpp , building the foo.i target will preprocess both files.
Specifying a Build ProgramВ¶
The program invoked by the —build mode is determined by the CMAKE_MAKE_PROGRAM variable. For most generators, the particular program does not need to be configured.