Lets define the n-th Fibonacci number fn as follows:
Program a template program in C++ that computes at compile time the n-th Fibonacci number and writes its value encoded in some compiler output message (e.g., error message) for all n from 1 to m for a fixed given m. The example prime.C might be of help.
What do you think of the compilation time with respect to the parameter m (big-O calculus)? What would be your computational model (assumptions about the compiler)?