Looking for an alternative to an enum (or any non trivial type)

Question

Hi. I am making a class to manage geometric points, as in [x, y, z]. some languages, such as scripting language 'adobe actionscript' support certain special values such as NaN (or 'not a number'), and I can do things such as
var _some_var:Number = Number.NaN;
the previous sentence is something like assigning null to a pointer. NaN is not a valid value, so I can not operate with a var that has nan as a value.
I want to do something like that in my class, but I can not use a number value and then say, 'this value is never gonna happen by itself, so if the var has that value is because is in nan state. for any value that I choose, if is a valid value, will be enough
to consider it as an invalid alternative. instead of that, the vars are handled in pairs, so for each numeric coordinate, such as x, there is also a bool that says if the numeric value is defined
in each point, there are the x, y and z coordinates, plus defx, defy and defz.
the constructor is something like
point() {
x= 0;
y= 0;
z= 0;
defx = false;
defy=false;
defz=false
and since all vars are private, I use accessors such as
setX(double _x) {
x = _x;
defX = true;
so how do I assign a NaN to a var that has already a value?, or, how do I set the var 'defX' to false again?
since setX(_some_value) is the same as x = _some_value, the idea would be that I could write something like
setX(NaN), and that would 'undefine' X.
since I can overload functions, and create functions with different prototypes for different types of values, even though the functions are called the same way, that's what I am doing.
I had two choices: the first one is (copy paste from my file)
//#ifndef _nan
//#define _nan
//#define NaN "NaN"
//#endif
not the one I chose because it does not seem better than my second choice (that's why it appears as a comment), or does not solve the problem I have with the choices I have thought so far:
(in file 'especiales.h')
#ifndef _especiales
#define _especiales
enum EEspecial {NaN};
#endif
and in the file for the point definition (as inline functions)
void setX(double _x) {
                x = _x;
                defX = true;
        void setX(EEspecial _e) {
                if (_e == NaN)
                    defX = false;
                else
                    throw coordIndefException();
        double getX();
        bool definidoX() {return defX;}
it works, it does not throw an error or there are no syntax problems; but there is a problem in my head. In the previous sentences there are two functions, one that takes a numeric value and sets it in x, and sets x as defined. on that receives not a numeric
value but a special value, that if it turn out to be nan, will unset the x var, and set it as undefined.I checked it with these simple lines:
    geometriaPunto _ensayo;
   try {
        _ensayo.setX(NaN);
        cout << _ensayo.getX() << "\n";
    } catch (exception _e) {
        cout << _e.what() << "\n";
this works fine. if I place a 0 in setX, it will print 0. if I place a NaN in the parenthesis, it will write the exception name. This works fine. But as far as I know, all basic var types, and all enums, also have implicit conversions to numeric values.
so even though it works here, as I understand, telling it setX(0) is the same as setX(NaN) (which would be the implicit conversion of NaN, since is the first value in the enum). none of those are double (because I did not wrote it as setX(0.0) ), so my guess
is that even though it works here, it was just luck, and some other c compile may understand it different, since there is not much difference between function_name(int) or function_name(enum_type)
what would be the best way, most clear way to overload 'setX()' so it does accept a type that can not be confused with a numeric value, yet it does not make my code way more obscure and does not create ambiguity?
tnx

Answer 1

this is the last post I will make on this thread, as I find your answers useless and I already have what I need, but concerning your answers:
"You said it worked for you in your 'adobe actionscript' code - why wouldn't the same technique work in C++ code? Number.NaN is a Number value, and it doesn't prevent you from doing operations in a var that holds that value - but that didn't
seem to concern you before."
have you ever worked with actionscript or did you just post because you had to post anything. yes, I can assign NaN to a numeric var in actionscript, just like I can assign null to a pointer. null is a value (0), if I add something to a var that has null,
it will not have null anymore, if the compiler or the executable don't verify this behavior that should throw an exception. In actionscript, if a var has nan and I add something to it, it will still have nan, because nan is not a value, is a state, and probably
actionscriopt also handles numeric vars with some other types of vars, to make these kind of validations. If I add 1 to garbage, I will still have garbage, and if I choose a random value to be chosen as nan, I or anybody else can not assure this value will
not be the result of a valid operation
2. "Is your question about C#, and not C++? If it is, you are in the wrong forum."
why do you ask, did you see the previous sentences written in c#, or do you still have to answer crap if you don't answer something related to the question. if somebody has been writing things in other languages (including java or the same actionscript,
which support that kind of behavior for properties and accessors), seems logic that I want to maintain that programing style, which will help somebody who also has been writing in those languages and then tries to read this new source, written in c++