While many simple programs fit into a single C or CPP source file, any serious project is going to need splitting up into several source files in order to be manageable. However, many beginning programmers may not realize what the point of this is - especially since some may have tried it themselves and run into so many problems that they decided it wasn't worth the effort. This article should explain why to do it, and how to do it properly. Where necessary, I give a brief explanation of how compilers and linkers work to help understand why you have to do things in a certain way.
In this article, I will call standard C and C++ files (usually with the .C or .CPP extension) "source files". This will be to distinguish them from "header files" (usually with the .H or .HPP extension). This terminology is also used by Visual C++ and most books. Note that the difference is purely conceptual - they are both just text files with code inside them. However, as the rest of the article will show, this difference in the way you treat source and header files is very important.
Why split code into several files?
The first question some novice programmers ask when they see a directory full of separate code files is, "why isn't it all just in one file?" To them, they just don't see the point of scattering the code about.
Splitting any reasonably-sized project up buys you some advantages, the most significant of which are the following:
- Speed up compilation - most compilers work on a file at a time. So if all your 10000 lines of code is in one file, and you change one line, then you have to recompile 10000 lines of code. On the other hand, if your 10000 lines of code are spread evenly across 10 files, then changing one line will only require 1000 lines of code to be recompiled. The 9000 lines in the other 9 files will not need recompiling. (Linking time is unaffected.)
- Increase organization - Splitting your code along logical lines will make it easier for you (and any other programmers on the project) to find functions, variables, struct/class declarations, and so on. Even with the ability to jump directly to a given identifier that is provided in many editors and development environments (such as Microsoft Visual C++), there will always be times when you need to scan the code manually to look for something. Just as splitting the code up reduces the amount of code you need to recompile, it also reduces the amount of code you need to read in order to find something. Imagine that you need to find a fix you made to the sound code a few weeks ago. If you have one large file called GAME.C, that's potentially a lot of searching. If you have several small files called GRAPHICS.C, MAINLOOP.C, SOUND.C, and INPUT.C, you know where to look, cutting your browsing time by 3/4.
- Facilitate code reuse - If your code is carefully split up into sections that operate largely independently of each other, this lets you use that code in another project, saving you a lot of rewriting later. There is a lot more to writing reusable code than just using a logical file organization, but without such an organization it is very difficult to know which parts of the code work together and which do not. Therefore putting subsystems and classes in a single file or carefully delineated set of files will help you later if you try to use that code in another project.
- Share code between projects - The principle here is the same as with the reuse issue. By carefully separating code into certain files, you make it possible for multiple projects to use some of the same code files without duplicating them. The benefit of sharing a code file between projects rather than just using copy-and-paste is that any bug fixes you make to that file or files from one project will affect the other project, so both projects can be sure of using the most up-to-date version.
- Split coding responsibilities among programmers - For really large projects, this is perhaps the main reason for separating code into multiple files. It isn't practical for more than one person to be making changes to a single file at any given time. Therefore you would need to use multiple files so that each programmer can be working on a separate part of the code without affecting the file that the other programmers are editing. Of course, there still have to be checks that 2 programmers don't try altering the same file; configuration management systems and version control systems such as CVS or MS SourceSafe help you here.
All of the above can be considered to be aspects of modularity, a key element of both structured and object-oriented design.
Next : How to do it