Inline function??

[atjm88]

Hi, can I know what's Inline function and how to use them?

#include <iostream>
#include <conio>

inline float cube(const float a) {return a*a*a;}

main(){
float side;
cout<<"Enter the side length of your cube: ";
cin>>side;
cout<<"Volumn of cube with side "<<side<<" is "<<cube(side)<<endl;
getch ();
return 0;
}

[hamsterman]

Keyword inline gives the compiler a hint that a certain kind of optimization, called inlining, might be a good idea. Whether to perform it is for the compiler to decide. See http://en.wikipedia.org/wiki/Inline_function

inline functions can (and should) be put into headers as the compiler needs to see the body of a function to inline it.

[atjm88]

inline functions can (and should) be put into headers as the compiler needs to see the body of a function to inline it.

1. which one is header?

2. and is this inline float cube(const float a) {return a*a*a;}
same as this

inline float cube(const float a)
{
	int r;
	r=a*a*a;
	return r;
}

3. why do we need to put "const" in line 4? Thanks...

[hamsterman]

1. header is a file where you hold declarations of classes and functions. If you're only working on small programs (one cpp file), you can ignore this, you'll learn later.

2. yes. edit: whoops.

3. we don't.

[Stewbond]

Here's a quick related question:

I've seen people use const like that. Is there any benefit to doing this other than just write protection?

[Gaminic]

@atjm88:
2. No! Line 4 will truncate a*a*a to an integer. Your compiler will probably generate a warning ("possible loss of precision"). The shorter version doesn't have this problem because it doesn't use an intermediate variable to store the result. If you declare 'r' as a float, the two code snippets will be identical in behaviour.

@Stewbond: Theoretically, there could be some optimizations through constness, but since const isn't a guarantee of constness, this is probably not the case. Personally, I find 'const' sends a clear message about the meaning. If a function shouldn't change an argument, then it shouldn't be able to.

(There's a post in the FAQ-lite about how you shouldn't const for performance. I'll look it up if I have a moment to spare.)

[atjm88]

@Gaminic, ya...u r right...I miss out the float...But my compiler(Borland C++) did not give any warning about it, it'll just shows the answer in int, that's why I miss out...Now problem solve with float...thanks...

but for the const...I still don't understand when we need it and when not...Because when I delete that const inside the code, it still works fine...

[atjm88]

One more question is, I just tested on VSC++ 2010, it can't works without int inside line 12 below...May I know why do we need "int" in some compiler and some no need? And can I use float main () instead of int main () in line 12?

Error shown when no int in line 12

error C4430: missing type specifier - int assumed. Note: C++ does not support default-int

#include <iostream>
#include <conio.h>
using namespace std;

inline float cube(float a)
{
	int r;
	r=a*a*a;
	return (r);
}

int main(){
float side;
cout<<"Enter the side length of your cube: ";
cin>>side;
cout<<"Volumn of cube with side "<<side<<" is "<<cube(side)<<endl;
getch ();
return 0;
}

[Stewbond]

main() is the first function called by your machine when your program is excecuted. It has to be of type int by definition. void used to be acceptable, but not anymore.

[hamsterman]

const is never really necessary (unless some other const enforces it). It's useful for optimization and documentation.