C++ temporaries

C++ temporaries

[a.biardi]

Hi all,

I have a function similar to the following, which returns a string:

  std::string foo() {
    return <whatever>;
  }

..then I want to use its c_str(), say like this (sorry for the silly 
example):

  int main() {
    printf("foo is %s",foo().c_str());
  }

Of course, for printf() to work, the temporary object that is returned 
by foo() should not be (automatically) destroyed too early, or the 
pointer returned by c_str() would not be valid anymore.

So, my question is: *when* is that temporary string created and 
destroyed? is it safe to assume that it exists until printf() 
returns, and thus that I can use its c_str() this way?

I've always thought that the temporary would be destroyed *before* 
calling printf() [ create a string, get the value for c_str(), 
destroy the string, invoke printf() ] thus actually passing an 
invalid pointer in this case, but to my surprise I put together some 
lines and gcc proved me wrong. Is it just gcc? Or is it a standard 
C++ behavior?

Thanks,
Andrea.

Re: C++ temporaries

[Benjamin Sobotta]

Hi!

Actually I'm not quite sure about this, but I think you can assume that
this always works. I think what gets actually passed to printf is a
pointer to the value returned by c_str() which in turn is stored in some
independent temporary memory location until printf is done. 
I use a lot of constructs like this and it always worked for me, (Intel
& GNU) however I might still be wrong. :)

Benjamin

On Wed, 2005-01-19 at 21:56 +0100, a.biardi@tiscali.it wrote:
> Hi all,
> 
> I have a function similar to the following, which returns a string:
> 
>   std::string foo() {
>     return <whatever>;
>   }
> 
> ..then I want to use its c_str(), say like this (sorry for the silly 
> example):
> 
>   int main() {
>     printf("foo is %s",foo().c_str());
>   }
> 
> Of course, for printf() to work, the temporary object that is returned 
> by foo() should not be (automatically) destroyed too early, or the 
> pointer returned by c_str() would not be valid anymore.
> 
> So, my question is: *when* is that temporary string created and 
> destroyed? is it safe to assume that it exists until printf() 
> returns, and thus that I can use its c_str() this way?
> 
> I've always thought that the temporary would be destroyed *before* 
> calling printf() [ create a string, get the value for c_str(), 
> destroy the string, invoke printf() ] thus actually passing an 
> invalid pointer in this case, but to my surprise I put together some 
> lines and gcc proved me wrong. Is it just gcc? Or is it a standard 
> C++ behavior?
> 
> Thanks,
> Andrea.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-c-programming" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
-- 
Benjamin Sobotta <mayday@gmx.net>