Andrew Hardy wrote: > It is my understanding that this will enable the linker to assosiate > function calls in the 'C' compiled object file with the 'C' functions in > the 'C++' compiled object file. It really just turns off "name mangling" which provides for type safe linking. If it is off then you have C linkage. If on, then C++ linkage. It does not prevent you from using C++ features which require the C++ support library. Some people have had success using a limited subset of C++ to do that however. > If I remove any C++ features from the function definitions linking > succeeds, but as soon as I introduce any C++ feature to the function body, > then compile succeeds, but linking fails. I undewrstood the idea of this > feature was to allow C++ code to provide a 'C' interface for C code to call > and pass agruments! Actually the idea is to allow a C++ program to link in C libraries. There are all of these nifty C libraries and this provides a way to use them from a C++ program. The reverse of trying to use a C++ library in a C program is a stretch. Think of static global initializers such as "X x;" which has a constructor. The C++ startup initializes that variable. But a C startup does not have any code to do that and it won't get initialized. So you can easily link in C libraries into your C++ program. But a library of C++ code has expectations beyond C which are not met by the C startup. Bob
