I was thinking about the language that I am implementing and I wondered what a full set of data movement/replication operations would be. In C++ we have: assignment operator copy constructor std::swap (you can partially specialize it, though I don't and I doubt that many do) C++ will also be getting a move from what I understand. So given two variables A and B, with starting values of a, b or - (uninitialized) what are the various operations? Here is what I have identified: assignment (a b -> b b) copy construction (- b -> b b) swap (a b -> b a) move (a b -> b -) move construction (- b -> b -) There could be more operators such as an assignment operator that changes it right operand instead of its left, but since they are symetrical to the above cases I ommited them. You could also make a case that a conditional swap would be a sixth operator, called sort. Then the concept can be extended to more arguments, and swap becomes rotate left and rotate right. However I am not going to worry about that kind of thing for a long time if ever.

Ignoring language and compiler specific issues, such as memory allocation, or references, or something more obscure...

You need assignment.

Everything else is defined from there:
Swap(x, y): T = X; X = Y, Y = T;
Move(X, Y): Y = X;
CopyC(X): this = X;
MoveC(X): this = X;

The reason things are so complex in C++ is because C++ is complex language. In most way, far too complex. All of these operations are needed just because of all the quirks in allocation, subclassing, various reference semantics, and so on...

Your best bet is to solve these problems, rather than copying the C++ mess.

First and by far one of the worst possible situations that C++ gives you is the unitialized state. Every language since then has dealt with that once and for all, especially when dealing with classes.

What is the domain/purpose of the language are you working on?

plenty of people are making languages that aren't like C++. I like having references, pointers, and values all be distinct. So I am making a language with a lot in common to C++ semantically. Syntactically my language is somewhat like C++ but java and C# are both more closer to C++ than my language. I have added a lot more stuff too, first class functions, multiple dispatch, and tuples. What separates my language from C++ is that I have brutally elegant and nearly orthagonal syntax. My parser is 650 lines of easy to read C++.

Oh and the domain is general purpose.