Creating Static and Dynamic Libraries for your C Programs

G'day world! This is my first post here on dev.to. I'm stoked to be here, and today, I'm going to talk about code reuse in the C/C++ world.

Bill Gates never actually said "Hire the laziest person for the job, because they will find the easiest way to do it", but the sentiment certainly rings true.

If you're a Civil Engineer, you can't reuse a pylon from a previous bridge in a new building, but a Software Engineer can reuse code from a previous project. This is a huge productivity win, and we should be leveraging it as hard as we can.

When dealing with C or C++, there are two general ways of reusing code between projects - static libraries and dynamic libraries.

Static Libraries

A Static Library is just a series of object files (.o files) that have been archived together. On Windows, you'll see them with .lib extensions, whereas on linux systems they have .a extensions.

For example, let's say we have a project called "Wave" that makes web requests in C. We have two .c files, waveRequest.c and waveResponse.c. In order to create and use a static library, we do the following:

> # Create the object files
> gcc -c -o waveRequest.o waveRequest.c
> gcc -c -o waveResponse.o waveResponse.c
> # Create the Static Libaray
> ar rcs libWave.a waveRequest.o waveResponse.o
> # compile your code with the library
> gcc -o main main.c -lWave # The compiler adds the "lib" and ".a" itself

Now, whenever you want to use your Wave library, you can copy the libWave.a and compile it into your project. This has a few advantages:

• We are reusing code
• We can write unit tests for our Wave Library and be confident that it will work everywhere
• If the library gets updates and is not backwards-compatible, the older version is still compiled into your code

However, there are some disadvantages to using static libraries:

• Copying the .a file around to different projects is kinda clunky and doesn't feel like a nice workflow
• If the library is updated and you want to move to the new version, you need to recompile your project
• You might not be using all the functionality of the library, so compiling all of it in with your code could be a waste of space

So what do we do if we don't want to compile in the static library to our own code?

Dynamic Libraries

While you have to compile in your static library, the dynamic library is stored in one place and loaded into multiple programs at runtime (think of common C libraries, like stdlib, stdio and assert). They typically have .so file extensions (or .dll in Windows) and don't require any extra compile flags for use.

Using our Wave example, and assuming we have a Wave.h file, you can create a dynamic library as follows:

> # Compile the object files
> gcc -fPIC -c -o waveRequest.o waveRequest.c
> gcc -fPIC -c -o waveResponse.o waveResponse.c
> # Create the dynamic library
> gcc -shared -o libWave.so waveRequest.o waveResponse.o
> # Copy the library to where it can be included
> cp libWave.so /usr/local/lib/libWave.so
> cp Wave.h /usr/local/include/libWave.h

Now, there's a bit going on here, so let's unpack it. Firstly, we are compiling our object files with the -fPIC flag, which means "Position Independent Code" - this will remove all absolute address references from the code and use relative addresses instead, allowing us to reuse the code across multiple projects without having to recompile it.

We are also copying our library and header files to a place where the compiler knows to look for library and header files. This might be different on your machine, for info on osx, look here and for linux you can use ldconf.

We can use our wave library in code as follows:

#include <Wave.h>

int main(int argc, char* argv) {
    wave_make_request("https://dev.to");
    return 0;
}

Now that we are using the dynamic library, we get the following benefits:

• Whenever our library is upgraded, we automatically get the new version
• We don't have to copy files between projects
• We don't have to remember to compile in the library, just include it! But there are some drawbacks too:
• If updates are made to the library that aren't backwards compatible, you will probably end up with a runtime error
• On lower memory machines, paging between the dynamic library and your application code might cause a hit to your app's performance.

So there's my summary of static and dynamic libraries! Hopefully you found this useful, and we can all continue to be lazy.

Cheers

Comments

[Falcão]

I just learned about this in college and it was really fun, one thing that helped me understand it better was that static libraries are .a for "archive" and dynamic libraries are .so for "shared object" :)