inline redundancy?

Hey fellas,

I've been reading this tutorial and the guy uses the inline keyword like this:

Code:

---

class Foo
{
public:
    inline bool      Bar()      { return FooBar; }
}

---

Now I'm not sure (that's why I'm asking) but I think I read somewhere that this was unecessary, because a function like this will automatically be considered for inlining (at the compilers discretion).

Is that correct?

As far as I know, any function defined inside of the class itself is automaticly inlined. I don't have the standard, so I can't tell you for sure. But my understanding of it is, if you define a function inside the class declaration itself, it's inlined.

Quzah.

Thanks that's what I thought. Of course there's no guarantee of inline right? It's a "request" and the compiler decides.

Do you really need to define inline in the interface at all ? I thought it was enough to define it in the implementation. (I know it's all in one in this case, but imagine it was not, would I need to use inline in both, the interface and the implementation?)

You need to have the implementation in the h file. So either put it in the class and it will be inlined ( hopefully ) or put it in the h file after the class with the inline keyword.

Quote:

---

Originally posted by HybridM
Thanks that's what I thought. Of course there's no guarantee of inline right? It's a "request" and the compiler decides.

---

Right. It's like the 'register' keyword. It tells the compler you'd like this to be in a register, but it doesn't force it there. (There has been huge arguments on this on boards in the past, but trust me it's the way it works. :P)

Quzah.

afaik you can force it to be inline somehow.

Quote:

---

Originally posted by darksaidin
afaik you can force it to be inline somehow.

---

Not according to the C++ standard.

Nothing can fully force a function to always be inlined; usually, though, the compiler honors your request, or at the very least throws a warning when it can't.

I think some MS compilers have some sort of __force_inline.

I'm not sure if it's MS or another company, or if the keyword I used was correct, but I've definately seen it before.

Yes, MS has that, but even that doesn't make it 100% for certain inlined. It merely says to the compiler: "I don't care what optimizations you are doing, or whether you think it should be inlined or not, but *IF* the function can be inlined, do it."

Some kinds of function calls can't be inlined (such as via a function pointer) so nothing can ALWAYS inline a function.

True, very true.

just curious what is an inline function?

An inline function is a function that isn't called by the program like a regular function. Instead whenever the compiler sees a call to the inline function, it replaces the call with the actual code body of the function, thus avoiding the function call overhead.
It's not always better and can sometimes be worse to inline rather than just call normally, though.

someblubb.h

Code:

---

class CBlubb {
  void Blubbba();
}

---

someblubb.cpp

Code:

---

inline void CBlubb::Blubbba() {
  Blob();
}

---

So that would inline Blubbba if possible and if not, give me a warning about it, right?

btw, is it right that I can't have inlines in an abstract baseclass (of course not for abstract functions, but for functions that use abstract ones) ? If I inline those, the compiler always tells me it can't find an implmentation for *functionname*.