Thread: What is the purpose of header files?

Well,
you've pretty much answered it already
it's primarily for modularised code, and code reuse

Splitting files into Declaration (Header) and Implementation (source) files is a common concept, used in both procedural and OOP
however, it comes into it's own with OOP, so I'll stick with that

One of the core OOP concepts is the ability to hide the internal "workings" of an object, and only show certain, outward facing functions

you have your main program,
and it uses some object
You don't need to know "how" that object actually works internally
You only need to know how to use it

Classic example is the standard string library
you know how to use the string library
but you don't need to know exactly how it works
(all the dynamic allocation and resizing of buffers, and whatnot, is completely hidden)
You just #include the declaration and off you go

Code re-use also becomes important here, as well
if you have a common object that is used in many of your files, why keep repeating the same code?

Bung the object in a linked library and #include the objects header file in your source files (so they know how to use the library)
All of your files, when calling the common object, re-use the same library

But the point remains valid that modulation can be done without headers. As much as I hate java, Sun does it very well. They use interfaces, which are like headers to hide implementation. And I don't (even in C and C++) use the headers to figure out how to use something, I use a documentation of it's API, especially when the headers are HUGE.

OpenGL will be my example. I just opened it, it's 3034 lines of code. I'd much rather use the index of my Bluebook to find parameters instead of find it in that mess. The declarations are grouped, but not in any order.

gcc won't allow it but cocoa will allow you to #include a header or the cocao file itself.

Why did Seis get banned?