Possible fixes to FunctionPointer PR (#28)

[bogdanm]

#28 introduced a number of questionable changes:

• the clear method in FunctionPointerBase and FunctionPointerBind. This was introduced because it looks like it might be needed in MINAR. If it turns out that it's not, clear needs to be removed.
• private virtual destructor in FunctionPointerBase (following the already present private constructor). With that in mind, It might be that FunctionPointerBind and FunctionPointerX should not inherit from FunctionPointerBase, but rather include it, using composition (and a proper name change). This isn't an error, just a possible improvement if it turns out to be the right idea.

[autopulated]

Can you assign an uninitialised object instead of clear()-ing?

If the destructor needs to be virtual, then it should be public, I think?

[bogdanm]

Yup, that's why I thought that composition might be the better option here. After all, if both your constructor and destructor are protected, you're not expecting to ever cast your reference to the base class (which is the intended behaviour here), so that's not really an is-a relationship.
About assigning an unitiliazed object, that'll work, but I don't think the code will look much better.

[autopulated]

Actually, how would virtual inheritance even work here. Won't we always pass these objects around by value?

[bogdanm]

It would technically work, except you're not even supposed to use the base class directly. Yet another reason to think that inheritance is the wrong mechanism. That was actually protected inheritance initially, which is a bit better, but I still think composition is preferable to that.

[bremoran]

Composition should be okay.