Hi. I am really loving how mixin's are implemented, but I was wondering, When a class includes a mixin and overrides one of its member functions, Can we call the mixin's function explicitly anyway so we can extend its implementation? If not, would it be hard to add or break anything to add?

For example:

mixin class foo
{
    void print(){output("Hello");}
};

class bar : foo
{
    void print()
    {
        foo::print();
        output(" world!");
    }
};

It would break what the mixin is

mixins are not inherited, as they are not real class declarations. They are more like macros to allow code reuse without having support for multiple inheritance in the language.

If you want to be able to call the mixin's implementation of a method, then the mixin's method should have a different name so it is included in your class.

Yea, I just thought it would be nice to have the option to extend a default implementation instead of just override it. I guess I will just name it something else if needed, but then I need proxy functions for the 'default' implementation to get overridden. To achive the same, this is now needed:

mixin class foo
{
    void printfoo(){output("Hello");}
    void print(){printfoo();}
};

class bar : foo
{
    void print()
    {
        printfoo();
        output(" world!");
    }
};

I guess its not a big deal, it just seems annoying to have another layer of indirection for the 'default' behavior in this case especially since I doubt Angelscript can inline and eliminate the call for the default (presumably more often called) case. I guess it bugs me because one of the things I love most about Angelscript is how few proxy functions I need to write to get things done. Those always seemed so wasteful and ulgy to me. Code duplication is best avoided.

I understand they are macros instead of real inheritance , but why could the scope resolution operator not be allowed for them? Basically it would 'rename' mixin functions that have been overridden or just give them a 2nd name they can be accessed by. Or do these overridden functions not get compiled at all encase they now produce compile errors that are being avoided by the override?

It would bring them one step closer to being as useful as multiple inheritance without having to support MI, As multiple mixins could even have there own default implementations of the same function and the resulting class could pick what ones to call and in what order.

I actually like how mixins work more then MI, Mainly because it avoids the headache of virtual base classes in diamond inheritance when you want one variable to be accessible by multiple 'middle' classes to be derived from in various ways by other classes.

Of course, I don't know how complex this would be to add. I just thought if it was simple enough and didn't break anything, it might be considered but if not I can do without no problem.

It's doable, but it would definitely not a trivial solution. I'll add a note on my to-do list to analyze this in detail for a possible future improvement.

Awesome. That is all I can ask.

Also noticed the lack of constructors/destructors in mixins.. But I can understand that could get messy very quickly if used. So I'll stick to just adding functions for those if needed.