Let's say I have a simple project like this:
find_package(SDL2 REQUIRED)
add_executable(test main.cpp)
target_link_libraries(test PUBLIC SDL2::SDL2 SDL2::SDL2main)
install(TARGETS test DESTINATION .)
If I build with CMAKE_BUILD_TYPE=Debug, I will find that the DLLs I depend on (sdl2d.dll in this case) will be copied correctly to the binary folder alongside my application.
However, in a CMAKE_BUILD_TYPE=Release build, this doesn't seem to be the case - forcing me to add directives like this throughout my project:
if (WIN32)
IF(CMAKE_BUILD_TYPE MATCHES Debug)
install(FILES ${CMAKE_CURRENT_BINARY_DIR}/SDL2d.dll ${CMAKE_CURRENT_BINARY_DIR}/zstdd.dll DESTINATION .)
else()
install(FILES ${CMAKE_CURRENT_BINARY_DIR}/SDL2.dll ${CMAKE_CURRENT_BINARY_DIR}/zstd.dll DESTINATION .)
ENDIF()
install(FILES ${install_dir}/bin/assimp-vc142-mtd.dll DESTINATION .)
endif()
This is especially messy now that I'm trying to integrate error-reporting tools (i.e. Sentry) that need DLL and PDB files in order to see stack traces, etc. Is there any reason this behaves as such on windows, or some way to force it to act more consistently so that my builds or install outputs are actually in a form I can consistently zip up and release?
Related
I have a top-level CMakeLists.txt which includes another third party project from a subdirectory, like
add_subdirectory(ext/third_party/cmake)
third_party contains a library target which I want to build, but I want to modify some properties and want to avoid to modify the original CMake file. I do not want to link some of my own targets to that library, I'd rather want that third party library to be build with some modified properties and then put it into a custom output directory. So I do
set_target_properties(libthird_party PROPERTIES
# some properties that successfully get applied here
LIBRARY_OUTPUT_DIRECTORY "/my/output/dir")
I can see that other properties are successfully applied and the build is modified to my needs correctly, but the generated output library is not put into the directory I set. What could be the reason for that?
If this is a totally wrong or bad approach please also feel free to propose a better approach for my goal.
Figured it out myself with help from the comments. I was trying to modify a static library target, which is not affected by LIBRARY_OUTPUT_DIRECTORY (which only applies to dynamic libraries) but which needs the setting ARCHIVE_OUTPUT_DIRECTORY. So the corrected call is
set_target_properties(libthird_party PROPERTIES
# some properties that successfully get applied here
ARCHIVE_OUTPUT_DIRECTORY "/my/output/dir")
You have to deal with [LIBRARY, ARCHIVE, EXECUTABLE] x [Single, Multi]config generator x [Unix, Windows]way.
note: On Windows everything (.dll, .exe) is on the same directory while on Unix you generally have a bin and a lib directories.
include(GNUInstallDirs)
# Single config (e.g. makefile, ninja)
set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/${CMAKE_INSTALL_BINDIR})
if(UNIX)
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/${CMAKE_INSTALL_LIBDIR})
set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/${CMAKE_INSTALL_LIBDIR})
else()
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/${CMAKE_INSTALL_BINDIR})
set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/${CMAKE_INSTALL_BINDIR})
endif()
# For multi-config build system (e.g. xcode, msvc, ninja-multiconfig)
foreach(OUTPUTCONFIG IN LISTS CMAKE_CONFIGURATION_TYPES)
string(TOUPPER ${OUTPUTCONFIG} OUTPUTCONFIG)
set(CMAKE_RUNTIME_OUTPUT_DIRECTORY_${OUTPUTCONFIG}
${CMAKE_BINARY_DIR}/${OUTPUTCONFIG}/${CMAKE_INSTALL_BINDIR})
if(UNIX)
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY_${OUTPUTCONFIG}
${CMAKE_BINARY_DIR}/${OUTPUTCONFIG}/${CMAKE_INSTALL_LIBDIR})
set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY_${OUTPUTCONFIG}
${CMAKE_BINARY_DIR}/${OUTPUTCONFIG}/${CMAKE_INSTALL_LIBDIR})
else()
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY_${OUTPUTCONFIG}
${CMAKE_BINARY_DIR}/${OUTPUTCONFIG}/${CMAKE_INSTALL_BINDIR})
set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY_${OUTPUTCONFIG}
${CMAKE_BINARY_DIR}/${OUTPUTCONFIG}/${CMAKE_INSTALL_BINDIR})
endif()
endforeach()
I had a project which uses CMake as build tool and made a simple template for me and my collegues to use. As I searched for best and easy to use practices online, I've came across different approaches to make a library.
In this template, I've listed header files and source files in two seperate variables, and I'm not passing the headers to add_library command - just sources. And then I use set_target_properties with PUBLIC_HEADER variable to give the header-file list.
So far it seems to work, but I wonder if I'm making thing unnecessarily complex. Some people online give header files to add_library command as well and doesn't even use set_target_properties and such.
In short:
should we include header files to add_library or should we not (as a best practice)? And impacts of the two usage.
what is purpose being served by adding headers in the add_library/add_executable? As they seem working even without it (seems forward declaration and symbols only). confirm on understanding please.
(Here is the template I'm talking about:)
cmake_minimum_required(VERSION 3.1.0)
project(lae CXX C)
set(CMAKE_CXX_STANDARD 14)
include_directories(
${CMAKE_CURRENT_SOURCE_DIR}
)
set(SOURCE_FILES
...
)
set(HEADER_FILES
...
)
set( PRIVATE_HEADER_FILES
...
)
add_library(${PROJECT_NAME} SHARED ${SOURCE_FILES} )
set( REQUIRED_LIBRARIES
...
)
target_link_libraries(${PROJECT_NAME} ${REQUIRED_LIBRARIES} )
SET_TARGET_PROPERTIES(
${PROJECT_NAME}
PROPERTIES
FRAMEWORK ON
SOVERSION 0
VERSION 0.1.0
PUBLIC_HEADER "${HEADER_FILES}"
PRIVATE_HEADER "${PRIVATE_HEADER_FILES}"
ARCHIVE_OUTPUT_DIRECTORY "lib"
LIBRARY_OUTPUT_DIRECTORY "lib"
OUTPUT_NAME ${PROJECT_NAME}
)
In our projects we use a "simple" way of yours - add_library with both headers and sources.
If you add only sources, then you won't see headers in IDE-generated project.
However, when installing, we have to do it like that, using two install commands:
install(TARGETS library_name
LIBRARY DESTINATION lib)
install(FILES ${PUBLIC_HEADERS}
DESTINATION include/library_name)
If you want to do it as a single command, you can use set_target_properties with PUBLIC_HEADER, as you suggested.
Then, this kind of install is possible:
install(TARGETS library_name
LIBRARY DESTINATION lib
PUBLIC_HEADER DESTINATION include/library_name)
Choose the one you like the most and stick to it.
When I use set(CMAKE_DEBUG_POSTFIX "d"), the build and install targets work as expected. But in the libfooTargets-debug.cmake file with the exported targets, there is a path to libfoo and not libfood.
I exported the targets like this:
install(TARGETS libfoo EXPORT libfoo-targets LIBRARY DESTINATION lib ARCHIVE DESTINATION lib RUNTIME DESTINATION bin)
install(EXPORT libfoo-targets FILE libfooTargets.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
which creates and installs libfooTargets.cmake and libfooTargets-debug.cmake when building in debug mode, and libfooTargets.cmake and libfooTargets-release.cmake when building in release mode.
Both libfooTargets-release.cmake and libfooTargets-debug.cmake reference the name without a postfix as:
list(APPEND _IMPORT_CHECK_FILES_FOR_libfoo "${_IMPORT_PREFIX}/lib/libfoo.lib" )
and thus a program linking against the debug target still uses the release-build library and I would need to install release and debug versions into different folders to be able to link against the debug target.
How can I get the exported targets to work with a debug postfix?
I could of course try to change the library name depending on CMAKE_RELEASE_TYPE or a CONFIGURATION generator expression, but this will probably break the multi-configuration features in MSVC and other IDEs supporting different targets and seems not to work in the sense of how the exported targets feature is meant to simplify and unify the build.
I suspect that the install(EXPORT ...) command somehow drops the CMAKE_DEBUG_POSTFIX or does not implement it for generating the libfooTargets-{release,debug}.cmake files, but possibly I overlooked how to make this variable visible to the generator of the exported targets or something like this.
All target code
cmake_minimum_required(VERSION 3.11.1)
project(foo)
include(CMakePackageConfigHelpers)
set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
set(CMAKE_DEBUG_POSTFIX "d")
# ...
add_library(libfoo STATIC somesource.cpp someheader.h)
target_include_directories(libfoo PUBLIC
$<INSTALL_INTERFACE:include>
$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
)
target_link_libraries(libfoo
somelibrary
)
target_include_directories (libfoo PUBLIC
somelibrary_header_dirs
)
install(TARGETS libfoo EXPORT libfoo-targets LIBRARY DESTINATION lib ARCHIVE DESTINATION lib RUNTIME DESTINATION bin)
install(EXPORT libfoo-targets FILE libfooTargets.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
configure_package_config_file(libfooConfig.cmake.in ${CMAKE_CURRENT_BINARY_DIR}/libfooConfig.cmake INSTALL_DESTINATION ${CMAKE_INSTALL_PREFIX})
install(FILES ${CMAKE_CURRENT_BINARY_DIR}/libfooConfig.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
install(DIRECTORY include/ DESTINATION include FILES_MATCHING PATTERN "*.h")
The platform is a Windows 10 with cmake 3.11.1 and MSVC 2015. Of course the most general solution is probably the best one.
According to the documentation of the install command, you need to reference the configuration that you are interested in:
[...] If a CONFIGURATIONS option is given then the file will only be installed when one of the named configurations is installed. Additionally, the generated import file will reference only the matching target configurations. [...]
So, you need to add the CONFIGURATIONS option in both install commands and duplicate the commands for each configuration you want to install and export.
I have the simplest possible c-library which builds and is packed using the following CMakeLists.txt:
cmake_minimum_required(VERSION 3.5)
project (libfoo C)
add_library(foo SHARED impl.c)
target_link_libraries(foo)
install(TARGETS foo LIBRARY DESTINATION lib/)
install(FILES public_header.h DESTINATION include/libfoo)
set(CPACK_GENERATOR "TGZ")
include(CPack)
Working example is located here: https://github.com/bjarkef/cmake-simple/tree/master/libfoo
I execute mkdir -p build; (cd build/; cmake ../; make all package;) to build a .tar.gz package with the compiled shared library along with its public header file. This is all working fine.
Now I wish to modify the CMakeLists.txt to create the FooConfig.cmake and FooConfigVersion.cmake files needed for CMake find_package in a different project to find the foo library. How do I do this?
I have discovered I should used the CMakePackageConfigHelpers: configure_package_config_file and write_basic_package_version_file, and I should create a FooLibraryConfig.cmake.in file. However I cannot figure out how to put it all together.
Note that it is important the the resulting .cmake files only contains relative paths.
I have cmake module included in the top level CmakeList.txt:
# Generate and install package config files
include(PackageConfigInstall)
Within the generic PackageConfigInstall.cmake file, the config files are created from the cmake.in files, and installed. This module can be reused for other packages.
include(CMakePackageConfigHelpers)
# Generate package config cmake files
set(${PACKAGE_NAME}_LIBRARY_NAME ${CMAKE_SHARED_LIBRARY_PREFIX}${PACKAGE_NAME}${CMAKE_STATIC_LIBRARY_SUFFIX})
configure_package_config_file(${PACKAGE_NAME}-config.cmake.in
${CMAKE_CURRENT_BINARY_DIR}/${PACKAGE_NAME}-config.cmake
INSTALL_DESTINATION ${CMAKE_INSTALL_DIR}/${PACKAGE_NAME}
PATH_VARS LIB_INSTALL_DIR INCLUDE_INSTALL_DIR APP_INCLUDE_INSTALL_DIR )
configure_file(${PACKAGE_NAME}-config-version.cmake.in ${CMAKE_CURRENT_BINARY_DIR}/${PACKAGE_NAME}-config-version.cmake #ONLY)
# Install package config cmake files
install(
FILES
${CMAKE_CURRENT_BINARY_DIR}/${PACKAGE_NAME}-config.cmake
${CMAKE_CURRENT_BINARY_DIR}/${PACKAGE_NAME}-config-version.cmake
DESTINATION
${CMAKE_INSTALL_DIR}/${PACKAGE_NAME}
COMPONENT
devel
)
You'll need a package file for your library, such as your_lib-config.cmake.in, which will become your_lib-config.cmake. This will contain the include and library variables that can be used.
get_filename_component(YOUR_LIB_CMAKE_DIR "${CMAKE_CURRENT_LIST_FILE}" PATH)
# flag required by CMakePackageConfigHelpers
#PACKAGE_INIT#
set_and_check(YOUR_LIB_INCLUDE_DIR #PACKAGE_YOUR_LIB_INCLUDE_INSTALL_DIR#/hal)
set_and_check(YOUR_LIB_LIBRARY #PACKAGE_LIB_INSTALL_DIR#/#CMAKE_STATIC_LIBRARY_PREFIX##PROJECT_NAME_LIB##CMAKE_STATIC_LIBRARY_SUFFIX#)
set_and_check(YOUR_LIB_LIBRARIES #PACKAGE_LIB_INSTALL_DIR#/#CMAKE_STATIC_LIBRARY_PREFIX##PROJECT_NAME_LIB##CMAKE_STATIC_LIBRARY_SUFFIX#)
You'll also want a config-version.cmake.in file like this:
set(PACKAGE_VERSION #PACKAGE_VERSION#)
# Check whether the requested PACKAGE_FIND_VERSION is compatible
if("${PACKAGE_VERSION}" VERSION_LESS "${PACKAGE_FIND_VERSION}")
set(PACKAGE_VERSION_COMPATIBLE FALSE)
else()
set(PACKAGE_VERSION_COMPATIBLE TRUE)
if ("${PACKAGE_VERSION}" VERSION_EQUAL "${PACKAGE_FIND_VERSION}")
set(PACKAGE_VERSION_EXACT TRUE)
endif()
endif()
There's quite a bit to the packaging scripts to get it all to work just right. I went through a lot of trial and error to finally get something that works on different targets (both linux server and embedded target). I might have left something out, so please just comment and I'll update answer.
I had a project which uses CMake as build tool and made a simple template for me and my collegues to use. As I searched for best and easy to use practices online, I've came across different approaches to make a library.
In this template, I've listed header files and source files in two seperate variables, and I'm not passing the headers to add_library command - just sources. And then I use set_target_properties with PUBLIC_HEADER variable to give the header-file list.
So far it seems to work, but I wonder if I'm making thing unnecessarily complex. Some people online give header files to add_library command as well and doesn't even use set_target_properties and such.
In short:
should we include header files to add_library or should we not (as a best practice)? And impacts of the two usage.
what is purpose being served by adding headers in the add_library/add_executable? As they seem working even without it (seems forward declaration and symbols only). confirm on understanding please.
(Here is the template I'm talking about:)
cmake_minimum_required(VERSION 3.1.0)
project(lae CXX C)
set(CMAKE_CXX_STANDARD 14)
include_directories(
${CMAKE_CURRENT_SOURCE_DIR}
)
set(SOURCE_FILES
...
)
set(HEADER_FILES
...
)
set( PRIVATE_HEADER_FILES
...
)
add_library(${PROJECT_NAME} SHARED ${SOURCE_FILES} )
set( REQUIRED_LIBRARIES
...
)
target_link_libraries(${PROJECT_NAME} ${REQUIRED_LIBRARIES} )
SET_TARGET_PROPERTIES(
${PROJECT_NAME}
PROPERTIES
FRAMEWORK ON
SOVERSION 0
VERSION 0.1.0
PUBLIC_HEADER "${HEADER_FILES}"
PRIVATE_HEADER "${PRIVATE_HEADER_FILES}"
ARCHIVE_OUTPUT_DIRECTORY "lib"
LIBRARY_OUTPUT_DIRECTORY "lib"
OUTPUT_NAME ${PROJECT_NAME}
)
In our projects we use a "simple" way of yours - add_library with both headers and sources.
If you add only sources, then you won't see headers in IDE-generated project.
However, when installing, we have to do it like that, using two install commands:
install(TARGETS library_name
LIBRARY DESTINATION lib)
install(FILES ${PUBLIC_HEADERS}
DESTINATION include/library_name)
If you want to do it as a single command, you can use set_target_properties with PUBLIC_HEADER, as you suggested.
Then, this kind of install is possible:
install(TARGETS library_name
LIBRARY DESTINATION lib
PUBLIC_HEADER DESTINATION include/library_name)
Choose the one you like the most and stick to it.