我们想组织一个这样的C++
项目:
project/
lib1/ (first library)
CMakeList.txt
src/
lib1.c
foo1.h
build/
test/ (tests)
CMakeList.txt
test1.c
test2.c
lib2/ (second library)
CMakeList.txt
src/
CMakeList.txt
os/ (OS dependent code)
CMakeList.txt
win32/
xxx.c (win32 implementation)
linux/
xxx.c (linux implementation)
lib2.c
foo2.h
build/
include/ (shared/public headers)
lib1/
lib.h (shared library header included from apps)
lib2/
lib.h (shared library header -"-)
请在CMakeLists.txt
使用lib2
时使用link1
时如何写lib2
CMakeList.txt
应该是可移植的(至少是Win32,Linux ......)?
更正:如果某些{{1}}个文件不在他们的位置,请假设如此。我可能忘记了。
答案 0 :(得分:2)
整个理念是从整个项目的中心CMakeLists.txt开始。在这个级别,所有目标(libs,可执行文件)都将被聚合,因此例如从lib1到lib2的链接没有问题。如果lib2将链接到lib1,则需要首先构建lib1。
平台特定的源文件应该有条件地设置为某个变量。
(如果需要在子目录中设置变量并在上面的目录中使用它,则必须使用CACHE FORCE等将其设置为缓存 - 请参阅set
的手册)
这就是你如何正确地完成源代码构建 - 正如CMake所希望的那样:
cd project-build
cmake ../project
每个库具有单独的构建目录并不是非常CMake'ish(如果我可以这样说)并且可能需要 一些黑客。
project-build/
project/
CMakeLists.txt (whole project CMakeLists.txt)
[
project(MyAwesomeProject)
include_directories(include) # allow lib1 and lib2 to include lib1/lib.h and lib2/lib.h
add_subdirectory(lib1) # this adds target lib1
add_subdirectory(lib2) # this adds target lib2
]
lib1/ (first library)
CMakeList.txt
[
add_library(lib1...)
add_subdirectory(test)
]
src/
lib1.c
foo1.h
test/ (tests)
CMakeList.txt
test1.c
test2.c
lib2/ (second library)
CMakeList.txt
[
add_subdirectory(src)
]
src/
CMakeList.txt
[
if(WIN32)
set(lib2_os_sources os/win32/xxx.c)
elsif(LINUX)
set(lib2_os_sources os/linux/xxx.c)
else()
message(FATAL_ERROR "Unsupported OS")
endif()
add_library(lib2 SHARED lib2.c ${lib2_os_sources})
]
os/ (OS dependent code)
win32/
xxx.c (win32 implementation)
linux/
xxx.c (linux implementation)
lib2.c
foo2.h
include/ (shared/public headers)
lib1/
lib.h (shared library header included from apps)
lib2/
lib.h (shared library header -"-)