prototype declaration not variable definition

[John Harrison]

I've seen cases before where it was intended to write a variable declaration
but a function prototype was written instead (e.g. X x();) but I've come
accross a new version of this (new to me at least) and I want to check I've
understood it correctly.

The code I wrote was.

std::istringstr eam buf("abc");
std::vector<cha r> vec(std::istrea m_iterator<char >(buf),
std::istream_it erator<char>()) ;

To my surprise vec was interpreted as a function prototype. After a bit of
thought here's how I see it

std::istream_it erator<char>(bu f)

is a parameter declaration with std::istream_it erator<char> as the type and
buf as the dummy parameter name, and also a superfluous pair of parens. So
far so good, the bit that I couldn't work out was how

std::istream_it erator<char>()

was being interpreted as a parameter declaration. Then it occured to me that
in function declarations you are allowed to omit the dummy parameter name,
so that std::istream_it erator<char>() could be interpreted as parameter
declaration with superfluous parens surrounding the missing dummy parameter
name!

Is this a correct interpretation? Or am I barking?

john

[Victor Bazarov]

Re: prototype declaration not variable definition

"John Harrison" <john_andronicu s@hotmail.com> wrote...[color=blue]
> I've seen cases before where it was intended to write a variable[/color]
declaration[color=blue]
> but a function prototype was written instead (e.g. X x();) but I've come
> accross a new version of this (new to me at least) and I want to check[/color]
I've[color=blue]
> understood it correctly.
>
> The code I wrote was.
>
> std::istringstr eam buf("abc");
> std::vector<cha r> vec(std::istrea m_iterator<char >(buf),
> std::istream_it erator<char>()) ;
>
> To my surprise vec was interpreted as a function prototype. After a bit of
> thought here's how I see it
>
> std::istream_it erator<char>(bu f)
>
> is a parameter declaration with std::istream_it erator<char> as the type[/color]
and[color=blue]
> buf as the dummy parameter name, and also a superfluous pair of parens. So
> far so good, the bit that I couldn't work out was how
>
> std::istream_it erator<char>()
>
> was being interpreted as a parameter declaration. Then it occured to me[/color]
that[color=blue]
> in function declarations you are allowed to omit the dummy parameter name,
> so that std::istream_it erator<char>() could be interpreted as parameter
> declaration with superfluous parens surrounding the missing dummy[/color]
parameter[color=blue]
> name!
>
> Is this a correct interpretation? Or am I barking?[/color]

You're absolutely correct.

Victor

[Buster Copley]

Re: prototype declaration not variable definition

John Harrison wrote:[color=blue]
> I've seen cases before where it was intended to write a variable declaration
> but a function prototype was written instead (e.g. X x();) but I've come
> accross a new version of this (new to me at least) and I want to check I've
> understood it correctly.
>
> The code I wrote was.
>
> std::istringstr eam buf("abc");
> std::vector<cha r> vec(std::istrea m_iterator<char >(buf),
> std::istream_it erator<char>()) ;
>
> To my surprise vec was interpreted as a function prototype. After a bit of
> thought here's how I see it
>
> std::istream_it erator<char>(bu f)
>
> is a parameter declaration with std::istream_it erator<char> as the type and
> buf as the dummy parameter name, and also a superfluous pair of parens. So
> far so good, the bit that I couldn't work out was how
>
> std::istream_it erator<char>()
>
> was being interpreted as a parameter declaration. Then it occured to me that
> in function declarations you are allowed to omit the dummy parameter name,
> so that std::istream_it erator<char>() could be interpreted as parameter
> declaration with superfluous parens surrounding the missing dummy parameter
> name!
>
> Is this a correct interpretation? Or am I barking?
> john[/color]

Close. The second parameter of vec is a pointer to a function taking no
arguments and returning std::istream_it erator <char>.

#include <sstream>
#include <vector>
#include <iterator>

std::istream_it erator <char> b;
std::istream_it erator <char> f ();

int main ()
{
std::istringstr eam buf ("abc");
std::vector <char> vec (std::istream_i terator <char> (buf),
std::istream_it erator <char> ());

sizeof vec (b, f); // use the declarations; no error
}

Regards,
Buster.

[John Harrison]

Re: prototype declaration not variable definition

> >[color=blue][color=green]
> > Is this a correct interpretation? Or am I barking?
> > john[/color]
>
> Close. The second parameter of vec is a pointer to a function taking no
> arguments and returning std::istream_it erator <char>.
>
> #include <sstream>
> #include <vector>
> #include <iterator>
>
> std::istream_it erator <char> b;
> std::istream_it erator <char> f ();
>
> int main ()
> {
> std::istringstr eam buf ("abc");
> std::vector <char> vec (std::istream_i terator <char> (buf),
> std::istream_it erator <char> ());
>
> sizeof vec (b, f); // use the declarations; no error
> }
>
> Regards,
> Buster.
>[/color]

That would be

std::vector <char> vec (std::istream_i terator <char> (buf),
std::istream_it erator <char> (*)());

wouldn't it?

john

[John Harrison]

Re: prototype declaration not variable definition


"Buster Copley" <buster@none.co m> wrote in message
news:bgso48$elf $1@newsg2.svr.p ol.co.uk...[color=blue][color=green][color=darkred]
> >>Close. The second parameter of vec is a pointer to a function taking no
> >>arguments and returning std::istream_it erator <char>.[/color][/color]
>[color=green]
> > That would be
> >
> > std::vector <char> vec (std::istream_i terator <char> (buf),
> > std::istream_it erator <char> (*)());
> >
> > wouldn't it?
> >
> > john[/color]
>
> Yes, that would work too.
> Buster
>[/color]

But my point was I didn't realise

void f(int ());

was a synonym for

void f(int (*)());

How long has that been part of the language?

And how is that interpretation of void f(int()) preferred over mine. Isn't
that another ambiguity?

john