"C4127: conditional expression is constant" Huh?

Quote:

---

Originally Posted by Elysia

It doesn't do that. Not in Visual Studio at the very least.

---

Too bad for Visual Studio then. Statements that have no effect are normally quite bad.

But then this "fix" isn't really portable. :)

Quote:

---

Originally Posted by anon

Too bad for Visual Studio then. Statements that have no effect are normally quite bad.

---

Are not. The compiler will ignore them anyway.
And it's a great way to remove the "unreferenced" warning.
I think it's a good thing.

Quote:

---

The compiler will ignore them anyway

---

That's the point. In most contexts you do not write a code to be ignored by the compiler. So when such a code is encountered - it is very likely to be an error

Ah yes, but they do work fine for things such as suppressing warnings, if the compiler supports it. I think it is a good feature, anyway.

I'm using following macro for unused function argumants - but I do not know if some compiler
will complain
(unused local vars I just remove or place inside proper #ifdefs)

Code:

---

#define UNUSED_ARG(_arg) if (_arg) {}

---

Wouldn't it be quite easy to do this?

Code:

---

void foo(int , int y) //first argument is unused
{
    //use y
}

---

Yep, that's what I do for function arguments.

Quote:

---

Originally Posted by anon

Wouldn't it be quite easy to do this?

---

I write mostly in C

I'll give you another example:

Code:

---

void func( char* str, int num )
{
    ASSERT( num < 50 );
...  // num isn't used anywhere else.
}

---

I encountered some 3rd party code similar to this. Since they use the ASSERT() macro, num is used in Debug mode but not in Release mode. God only knows what the point is of this, but I didn't write it and I don't want to risk breaking anything, so the only ways of silencing the warning about unused parameter is either use a #pragma, or add num; to make the compiler shut up.

#pragma statements (and warning numbers/messages) are different between compilers, so I chose num; instead.

>Statements that have no effect are normally quite bad.
Agreed.

To do VC++ justice, it seems that only expressions of type p; don't cause a warning. If there's an operator or such thrown into the mix, you get the "effectless statement" warning (e.g a + b; ).

I could still imagine such a situation where a warning might draw attention to a problem:

Code:

---

//global
int c;

void foo()
{
    int a, b; c; //one semicolon by mistake
    ...
}

---

I don't think I've ever done such a mistake...
But then, if you tried to use c, you'd get an undeclared warning and it should be pretty obvious then that you have done a mistake there...

I think I have made such mistakes (semicolon instead of comma), and in this example there is a global int c. Meaning the whole function will use this, instead of a local c that was supposed to shadow it.

It is a contrived example, though...