Variable declaration taken as a function pointer declaration
Question posted by: Bolin
(Guest)
on
December 2nd, 2005 04:05 PM
When compiling the following code:
[code]
#include <iostream>
struct B {};
struct A
{
A(B b1, B b2) {};
void foo() { std::cout << "foo called" << std::endl; }
};
int main(int argc, char * argv[])
{
A a(B(), B());
a.foo();
return 0;
}
[\code]
the compiler will interprete the first line of the main function as a
function pointer declaration, and thus will fail at the next line. Does
somebody know why this is so, and if there is an elegant way to solve
this problem that does not involve temporary variables?
B.
4
Answers Posted
Bolin wrote:[color=blue]
> When compiling the following code:
>
> [code]
> #include <iostream>
>
> struct B {};
>
> struct A
> {
> A(B b1, B b2) {};[/color]
.. ^^^
Trailing semicolon is extraneous here.
[color=blue]
> void foo() { std::cout << "foo called" << std::endl; }
> };
>
> int main(int argc, char * argv[])
> {
> A a(B(), B());
> a.foo();
> return 0;
> }
> [\code]
>
> the compiler will interprete the first line of the main function as a
> function pointer declaration, and thus will fail at the next line. Does
> somebody know why this is so, and if there is an elegant way to solve
> this problem that does not involve temporary variables?[/color]
It is so because the language designers had to make a decision and they
picked the "If it looks like a declaration, it is a declaration" solution.
You can work around it by adding an extra set of parentheses around the
arguments:
A a((B()), (B()));
V
Bolin wrote:[color=blue]
> When compiling the following code:
>
> [code]
> #include <iostream>
>
> struct B {};
>
> struct A
> {
> A(B b1, B b2) {};
> void foo() { std::cout << "foo called" << std::endl; }
> };
>
> int main(int argc, char * argv[])
> {
> A a(B(), B());
> a.foo();
> return 0;
> }
> [\code]
>
> the compiler will interprete the first line of the main function as a
> function pointer declaration, and thus will fail at the next line. Does
> somebody know why this is so, and if there is an elegant way to solve
> this problem that does not involve temporary variables?[/color]
It is so because your code could either be read as a function
declaration or a variable definition and the rule to disambiguate this
is "if it could be a function declaration, it is a function
declaration". The solution is to change the line that defines a to
A a((B()), (B()));
Gavin Deane
Thanks for your answer and your solution -- the thing that surprised me
is that a bare B() could be interpreted as a function pointer.
B.
Bolin wrote:[color=blue]
> Thanks for your answer and your solution -- the thing that surprised me
> is that a bare B() could be interpreted as a function pointer.[/color]
Take a look at
http://www.gotw.ca/gotw/075.htm
particularly the logic up to examples 2d and 2e.
Gavin Deane
|
|
|
What is Bytes?
We are a network of experts and professionals in IT and software development that help one another with answers to tough questions and share insights.
Get the best answers to your questions from over 196,944 network members.
Top Community Contributors
|