Circ Dependency hell [Solved]

example header

main.h


now you can include main.h as many times as you want



You just need to use compiler directives so that it won't include the same content twice

Whats happening?

#ifndef is pretty much 'if not defined', so in this case its looking for token GUARDMAIN_H, if its not found then it proceeds with reading and compiling the code - its only processed during compile time (I think they're called Compiler Directives)

#define GUARDMAIN_H - just defining the token

#endif - and ending the #ifndef

I think you could also use #pragma once - but I'm not too sure about that

*edit* Yeah you can use #pragma once, it seems it has had some history!

Lol, was not expecting this for a response. I'm well aware of include guards but didn't include(lol) them for the sake of brevity. I get a C2079, not C2011, which I had originally included in the title, but decided against obscure numbers.

Ahh sorry about that,

I think you can do an extern class b //edit just tested and doesn't seem to be the case.

Though if not, you will just have to include b.h in c.h also - if I read right.

eta: Why would c++ have any form of brevity? :p

Extern creates a static instance, not what i'm looking for, and i don't think it will help anyway. Generally I don't post until I've exhausted all other options, and b in c didn't work, nor any other permutation i've thought of.

Give this link a try http://msdn.microsoft.com/en-us/library/9ekhdcxs%28VS.80%29.aspx

I've hit that link already,



even though your suggestions as yet haven't helped, I very much appreciate them. Keep them coming;

You can use a forward declaration for references.

With FD's only and NOT including C.h in b.h i get :

b.h(xxx) : error 2027: use of undefined type 'c'b.h(xxx) : error C2228: left of '.c_member_property' must have class/struct/union

Is the issue that I'm trying to use member data from C in a function of b? these are all low level utility classes with no private data of there own.

I could always make a global function:
BFunctionThatDependsOnInstanceOfC( C _c) but have been trying to avoid this as much as possible.

Well, the obvious solution would be to move the function definition from the header to a source file.

You cannot call member functions of forward declared classes. The compiler must see the function declarations to preserve the type system and generate the correct code.

Can you move the implementation into a source file?

Quote:

Original post by rip-off You cannot call member functions of forward declared classes. The compiler must see the function declarations to preserve the type system and generate the correct code. Can you move the implementation into a source file?

???

All member functions are implemented in their respective .cpp files, is this what you meant? If not please explain...

Then why is your error message listing the header as the location of the error?

@SiCrane: Typo...

b.cpp(xxx) : error 2027: use of undefined type 'c'    b.h(xxx) : see declaration of 'c'b.cpp(xxx) : error C2228: left of '.c_member_property' must have class/struct/union

@DimitriA: I apologize, your link I believe is the solution after all...

Each class, a,b,and c contain unnamed unions, which I now believe to be the cause of all my frustration... suggestions still welcome.

Sounds like you haven't included c.h in b.cpp.

Thanks for the help all... consider it solved issue was that class C had a member which was a union with a struct containing class A. Not an issue until C was needed in B.

Well... What I had thought was solved, was not, but is now!

The issue was the inclusion of needed headers in the .h files. ThrustGoblin's post nailed it on the head. moving the #include directives into the .cpp files was the fix i wanted. I'm surprised that this is the first I've ran into it, but this was the first case of 3 interdependent classes I've come across. Up until now I had always used #include in the .h files, which was fine(with forward declarations) for 2 interdependent classes, but with 3 the compiler could never build a concrete definition of the classes. I see the folly of my ways, and would like to thank all again for the help, and allowing me to better understand the c++ language.