wampirr / sample_cmake

This repo is to understand the working of cmake and make in a C++ project

sample_cmake

This repo is to understand the working of cmake and make in a C++ project

---

What is CMake?

It is a build tool that basically takes in the name of the source file(s) and various parameters mentioned in a file called CMakeLists.txt and spits out a make file that in turn could be used to create the final executable binary.

Initial file structure

• CMakeLists.txt (only file required by cmake)
• main.cpp (source file)

Build process

The usual trend is to create a build directory so as to hold the output of cmake in a seperate directory. Then we point cmake to the location where CMakeLists.txt resides. Then CMake shall spit out several files out of which make is the one we are concerned with. Once cmake has generated the files, there is no need to run cmake again in case any changes are made to CMakeLists.txt; make will know about these changes automatically.

1. mkdir build
2. cd build
3. cmake ..
4. make ( or make VERBOSE=1 to see under the hood )

This shall generate the binary executable in the build directory that can be run to verify that the make file indeed compiles the project

Note: Clean the build directory and try the above commands

Final file structure

• CMakeLists.txt (only file required by cmake)
• main.cpp (source file)
• build/
  • CMakeFiles/
  • cmake_install.cmake
  • CMakeCache.txt
  • Makefile
  • output

---

What is Make?

The main task of Make is to speed up the build process when we have to compile many files at ones. CMake after execution generates a make file. A make file is nothing but a file holding raw commands that need to be executed in order to build the binary. There can be multiple targets in a make file, but by default, it will execute the first one.

Note: Writing Makefile is a tedious process, hence CMake is a utility to automate that task.

Structure of a make file

A Makefile will have the following structe:

target_name_1:
	g++ a.cpp b.cpp c.cpp -o output
target_name_2:
	gcc a.cpp b.cpp -o output

To execute a particular target:

make target_name

A rather comlex example of make ( where one target can call the other )

target: dependency1 dependency 2
command_depending_on_the_dependencies

CC=g++

CFLAG=-c -Wall

all: hello

hello: main.o function1.o function2.o
	$(CC) main.o function1.o function2.o -o output

main.o: main.cpp
	$(CC) $(CFLAG) main.cpp

function1.o: function1.cpp
	$(CC) $(CFLAG) function1.cpp

function2.o: function2.cpp
	$(CC) $(CFLAG) function2.cpp

clean:
	rm -rf *.o output

Note: Only files that have been changed after last compilation will be recompiled.

Manual compilation

The g++ compiler can be manually invoked to compile the project and to generate the binary like so:

g++ ../src/game_engine.cpp ../src/game_interface.cpp ../src/main.cpp -I ../src -I ../include -I ../libs/Logger/include -I ../libs/Randomize/include -L ../libs/Logger -l logger -L ../libs/Randomize -l randomize

Here -I flag is used to tell the compiler which directories contain the header files. -L flag is used to tell the compiler which directories contain the static library files (.a). -l flag is used to specify the name of the complier, which internally is prepended or postpended by the compiler. Hence here the compiler is looking for liblogger.a or loggerlib.a.

To convert the .o file(s) into .a file, we use the following command:

g++ -c .src/randomize_util.cpp -I ./include

Once this has been done, we use the ar tool to generate the archieve like so:

ar rsv librandomize.a a.o b.o c.o