subroutine in C

Quote:

---

Originally Posted by Elysia

Sure, if you only provide a definition, then you must use void in the parameter list.

---

Not so. In a definition, an unnamed void argument, and an empty argument list both mean the same thing.

So this will not compile with a "too many arguments" error:

Code:

---

int func(){
  return 0;
}

int main(){
  func(1);
}

---

Quote:

---

Originally Posted by King Mir

Not so. In a definition, an unnamed void argument, and an empty argument list both mean the same thing.

So this will not compile with a "too many arguments" error:

---

Not in VS2005

Quote:

---

Originally Posted by vart

Not in VS2005

---

Are you sure you did not compile as C++? It works for me when compiling as C.

EDIT:
Wait a minute, I assumed that King Mir meant 'this will not (compile with a "too many arguments" error)' (which does not really make sense, I suppose), but maybe King Mir meant 'this will (not compile) with a "too many arguments" error', in which case the reason why this is not so is that the definition is also a declaration.

EDIT #2:
hmm... but the language of C99 seems to indicate otherwise:

Quote:

---

Originally Posted by Section 6.7.5.3, Paragraph 14

An identifier list declares only the identifiers of the parameters of the function. An empty list in a function declarator that is part of a definition of that function specifies that the function has no parameters. The empty list in a function declarator that is not part of a definition of that function specifies that no information about the number or types of the parameters is supplied.

---

The wording "not part of a definition" seems to say that although a definition is a declaration, the fact that it is a definition trumps the fact that it is a declaration for the interpretation of an empty parameter list. If this is so (and also in C90), then gcc and MSVC are both non-conformant in this area.

I compiled it as C

VS2005 gives no errors
gcc port used in CodeBlocks - gives the error specified by King Mir

Quote:

---

Originally Posted by vart

gcc port used in CodeBlocks - gives the error specified by King Mir

---

What is the version of gcc? The MinGW port of gcc 3.4.5 did not emit such an error for both C90 and C99 mode with -Wall and -pedantic tacked on.

Quote:

---

Originally Posted by laserlight

What is the version of gcc? The MinGW port of gcc 3.4.5 did not emit such an error for both C90 and C99 mode with -Wall and -pedantic tacked on.

---

ok I played a little with the CodeBlocks installation to see exactly which gcc is configured, what switches are set, and finally - what command line it uses to compile a file.


Here is it:
1. gcc is 3.4.5 and it does not show the error. (I have running it from command prompt)
2. CodeBlocks on some reason starts g++ to compile the c-file


So - about the second - how do I tell to CodeBlocks that *.c files should be compiled as C and not C++?

Quote:

---

Originally Posted by vart

So - about the second - how do I tell to CodeBlocks that *.c files should be compiled as C and not C++?

---

I think this might work: click on the .c file and select "Properties...". Click on the "Advanced" tab and change "CPP" to "CC" for the "Compiler variable" setting.

EDIT:
No, Code::Blocks does not seem smart enough to use gcc instead of g++ in that case.

Quote:

---

Originally Posted by laserlight

The wording "not part of a definition" seems to say that although a definition is a declaration, the fact that it is a definition trumps the fact that it is a declaration for the interpretation of an empty parameter list. If this is so (and also in C90), then gcc and MSVC are both non-conformant in this area.

---

My take on the whole was:

Code:

---

void fun(void); // No parameters
void fun(); // No parameter information (infinite parameters)
void fun(void) { } // No parameters
void fun () { } // No parameters

---

I could be wrong, though.
Although VS would let the last example compile.