Home > Back-end >  How does compiler link in libraries in your code?
How does compiler link in libraries in your code?

Time:08-17

So I'm going through CS50 introduction course.

And I'm confused about the fact that if you write in IDE #include some header file, which tells computer to find some library, and your compiler will find that code and combine it with your code.

But then how does compiler find that code? Like I just type in #include <cs50.h> for example. But how does it find that code when I don't have it on my PC? Why would I have it without finding it online and downloading it beforehand? Does it look online and then download file, which it uses now and in future for my programs when I call #include? And if so does it mean that you can't properly program without connection to internet?

CodePudding user response:

And im confused about the fact that if you write in ide #include some header file which tells computer to find some library,

Be careful with terminology. #include tells the compiler to find some header. The word library in C usually refers to the file with compiled code (.a, .lib, .so, .dll).

and your compiler will find that code and combine it with your code.

This is right. By and large, the effect is the same as copying and pasting the contents of the header in the place of the #include statement.

But then how does compiler find that code? Like i just type in #include Cs50 for example.

The compiler has default search paths built in, where it looks for the header that is being #included. Typically directories like /usr/include and /usr/local/include are built into the compiler, as well as the current directory .. You can add more search paths through compiler command line arguments, for example -I/some/path in gcc and clang.

But how does it find that code when i dont have it on my pc

It doesn't. You will get an error like "Cs50: No such file or directory".

If this works on your university's system, probably the header is installed on that system in some central location.

why would i have it without finding it online and downloading it beforehand? Does it look online and then download file which it uses now and in future for my programs when i call #include?

It doesn't.

And if so does it mean that you cant properly program without connection to internet?

C was developed in the 1970s. The internet barely existed yet. You can program perfectly well in C without an internet connection – if you can do it without StackOverflow, of course ;)

CodePudding user response:

The code of the library must be present on your computer or nothing will work. There's no such thing as magic downloads of libraries, C compilers and linkers have worked the same since long before the Internet was even invented.

Standard library headers are typically downloaded & installed along with the compiler. They are also very likely already pre-linked into some convenient format for the target system. You don't need to worry about manually adding standard libs to your project since the compiler will take care of that for you.

Custom headers require their corresponding .c files or linked libs to be present too, but they must be manually added to the current project. Either by adding them in your IDE's project, or as in the old days by creating a make file.

How that works in CS50 I don't know, but the lib obviously comes pre-installed somehow and they have hidden how to the students. We wouldn't want to risk CS50 students actually learning how programming works, now would we...

CodePudding user response:

if you write in ide #include some header file which tells computer to find some library

Slight misunderstanding. When you type #include "somefile.h" into your program, the first stage of C and C compilation called the pre-processor will search for that file (called a "header") from the INCLUDE path. That is, the pre-processor will search through the local directory of the .c file, then a specified set of standard directories, and perhaps your own project directories to find a file called "somefile.h". The INCLUDE path is highly configurable with IDEs, command lines, and environment variables. Upon finding that file, the result is that the contents of that file are virtually substituted directly into your source code, exactly where the #include statement originally appeared. It's as if you had typed that exact file contents yourself into your .c file. After the textual substitution of the #include statements with the file contents, the intermediate file contents are handed off to the compiler stage to convert to object (assembly) code.

Like i just type in #include Cs50 for example. But how does it find that code when i dont have it on my pc

If the file can't be found, the pre-processor stage of the compile will fail and the whole thing comes to an end. I'm not sure what's in Cs50, but if you type #include "Cs50" and it works, then my guess is that your university environment has some project or environment configuration that adds a course specific include directory into your compilation path. Difficult to say since you didn't specify how you were building your code. Most IDEs will let you right-click on a #include statement and navigate to the actual source of the header file so you can inspect its contents and see where it originates from on your disk.

Also, your title says "linking", but you really mean "including". Linking is final stage after the compiler compiles all source files to object code to produce a final executable program.

CodePudding user response:

.h files (should only) contain data type definitions, extern declarations, and function prototypes. .h files are not libraries.

So the compiler will know what parameters function take and what is the return type. The compiler will know how to call them. It will also know the type of variables defined somewhere else in the code (including the libraries). When the compiler compiles the code it generates intermediate files called object files.

Those files are later linked together by a special program called linker. This program will find the actual code of the functions in other object files or libraries (Libraries are basically sets of object files grouped in one larger file). It happens behind the scenes. If you need to tell the linker to use specific library you simply use command line option. For example to use math library you need to use -lm compiler command line option.

But how does it find that code when I don't have it on my PC?

The library file has to be present in your file system. Otherwise linker will not be able to link functions or variables from that library.

And if so does it mean that you can't properly program without connection to internet?

No, it means that you have to have properly configured toolchain (ie all libraries needed present in your file system).

But then how does compiler find that code? For your level of knowledge: some libraries are linked by default. Other not - so you need to tell the compiler/linker what you want to use.

--gcc & binututils related--

Linking is generally quite a complicated process and the compiler has its own configuration files called "spec files" and linker has "linker scripts". But explaining what those files do is rather an advanced topic far beyond the scope of this question.

  • Related