February | 2019 | My Technical Blog

How do you use cmake in a project? Suppose the project is named as “myllvm” and has only one file: log.c and you have decided to use “gcc” to compile the C file.

This is the simplest CMakeLists.txt to get started:

The directory only have two files: log.c and CMakeLists.txt

To build the binary: either clang or gcc can be used, how to configure that?

First “mkdir build” and “cd build” and then issue the following command to use gcc to compile:

Otherwise, you can use clang (ie, replace the “gcc” with “clang”).

But it is also possible to modifiy CMakeLists.txt to have “gcc” specified inside.

How to add new directories with new C files? (the subdirectory is call “core”):

add_subdirectory(core)

How to add header directory into the compilation?

include_directories(include)

How to add multiple C or CPP files in “src” subdirectory?

#set(SOURCES src/mainapp.cpp src/Student.cpp)

#or just a widcard:

file(GLOB SOURCES “src/*.cpp”)

For more complete list of commands (for CMakeLists.txt) we can dig into cmake source codes and looking into “Help/command” directory. Or the commands are listed here:

And when you do a “cmake -G <generator>“ these are the generate currently available (under Help/generator directory):

How to make use of the available cmake recipes written for each different platform:

For example here some Linux related cmake recipes:

And here are the Windows related cmake templates:

How to use the cmake recipes for different common tasks (eg, finding nvidia cuda compiler, find header files, etc) which are called Modules.

For example of a subtasks, these are all the “finding compiler” related modules (left hand side number is the number of lines for the file – a good representation of the complexity of the tasks it is handling):