Function pointers vs. Derived classes

Question

So I have a Signal class and a client class which manages a list of Signal objects. The class provides a bunch of interfaces, one of them is an update()-function.

The update()-code is completely different for each signal. Thus, one would assume that's a job for derivation.

However, the client manages tens, maybe even a hundred different Signal objects. There is no way I want to have these many child classes.

I'm thinking of introducing a member updateFunction, which is a function pointer type. Now, the client could set this function (maybe via constructor) during the Signal's creation process. Then, the client can call update() for each Signal object and the update()-function would call the function pointer internally.

Would do you suggest? Is there a pattern for this?

Answer 1

If you have multiple types of Signals that have to be updated in different ways, then the cleanest solution is to create a sub-class for each type of Signal, where the sub-classes only implement the different update behaviours.

The proposal for storing a function pointer in the Signal class is just emulating manually what the compiler does for you when you create a sub-class with the added downside that the maintenance programmers (including you in a few months time) will have a harder time understanding the code.