Operator overloading syntax

[dave74]

I am looking at a piece of code that overloads an operator, the example I have compiles and works fine, however I don't quite understand it

//definition
date operator+( const date & lhs, int rhs ); 
bool operator==(const date &, const date &); //this I don't get

//implementation
date operator+( int lhs, const date & rhs )
{
	return rhs + lhs;
}

bool operator==(const date & lhs, const date & rhs)
{
    return lhs.compare(rhs) == 0;
}

When I get to the implementation I can follow it again. It's the declaration I am have trouble with.

const date & lhs

I follow, but

 const date &

I don't. I was expecting a name like lhs to be present??

Any explaination would be appreciated

[JaWiB]

Forward declarations do not require variable names for the argument:

int someFunc(int, int);

That will work, but in the implementation you must have a variable name:

int someFunc(int a, int b)
{
//...

}

[dave74]

Forward declarations do not require variable names for the argument

I didn't know that, thankyou.

The example I was looking at did some with variable names and some without.

Do you think it is better to stick to one way, and if so which one?

[Lurker]

Complete preference. Myself, I like to put in the names.

[lyx]

I think putting the names are useful in complex functions (look at the Windows API if you don't see what I mean -_-) where types aren't sufficient to tell what's the parameter role.
Well, if you don't know, just stick with the full qualification stuff. But, for an operator, the arguments are predefined by the langage (except for the () op.) so there's generally no need for names.

[CornedBee]

Putting in the names is also useful for helper tools like VC++'s IntellliSense, which presents you with the declaration of a function when you open the call parenthesis.

[bennyandthejets]

in the implementation you must have a variable name

Not true. If you do not use the variable in the function, it is not necessary to name it. For example:

int WINAPI WinMain(HINSTANCE,HINSTANCE,LPSTR,int)
{
return 0;
}

This is perfectly valid.

Personally, I don't specify variable names in funtion prototypes, because if for some reason I later decide to change the variable name, I must change it in two spots. Not a big effort I suppose, but to me it just seems the right way to do things. If you want to know each parameter's function by looking at the prototype, tack a comment on the end.

[Cat]

Originally posted by bennyandthejets
Not true. If you do not use the variable in the function, it is not necessary to name it. For example:

int WINAPI WinMain(HINSTANCE,HINSTANCE,LPSTR,int)
{
return 0;
}

This is perfectly valid.

Personally, I don't specify variable names in funtion prototypes, because if for some reason I later decide to change the variable name, I must change it in two spots. Not a big effort I suppose, but to me it just seems the right way to do things. If you want to know each parameter's function by looking at the prototype, tack a comment on the end.

Not true, it's perfectly legal to specify DIFFERENT names. Only the one in the definition matters; the name in the declaration is unused. And sometimes I DO make them different -- I want the names to be maximally descriptive in the declarations, but sometimes I don't use the same names (for various reasons) in the definition.