Why is there a need for reference in C++?

[sarathy]

Hi,
I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

- Is there anything that a reference can do that a pointer cannot ???

Please clarify,
Sarathy

[W Marsh]

On 29 Jul 2006 11:39:53 -0700, "sarathy" <sp*********@gm ail.com>
wrote:

>Hi,
I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

- Is there anything that a reference can do that a pointer cannot ???

Please clarify,
Sarathy

Pointers cause excess friction, and are believed to have been the
cause of explosions in space craft. References were invented for safe
use aboard spacebound vehicles.

[Kai-Uwe Bux]

sarathy wrote:

Hi,
I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

It wasn't fine, was it? Pointers *are* tricky. Now, in C++ the trickyness of
pointers is compounded by exceptions: since exceptions can divert the flow
of control at almost any point and toward unknown locations, the basic
requirement of a pointer that every new() is matched by a delete() along
each path of execution is harder to match. Thus, a device was created to
eliminate some uses of pointers. References and standard containers are in
this category. Other devices, like auto_ptr, were introduced to mitigate
the dangers for the remaining cases of pointer use.

- Is there anything that a reference can do that a pointer cannot ???

References can extend the life-time of temporaries:

#include <iostream>

struct log {

log ( void ) {
std::cout << "constructi on" << std::endl;
}

log ( log const & ) {
std::cout << "copy" << std::endl;
}

~log ( void ) {
std::cout << "destructio n" << std::endl;
}

void access ( void ) const {
std::cout << "access" << std::endl;
}

};

log create_tmp ( void ) {
return ( log() );
}

int main ( void ) {
{
log const & ref = create_tmp();
ref.access();
}
std::cout << std::endl;
{
// warning: UB
log const * ptr = &create_tmp( );
ptr->access();
}
}

// end of file
Best

Kai-Uwe Bux

[Daniel T.]

In article <11************ **********@75g2 000cwc.googlegr oups.com>,
"sarathy" <sp*********@gm ail.comwrote:

Hi,
I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

I read "The Design and Evolution of C++" by Bjarne Stroustrup a long
time ago so I may not remember this correctly, however as I remember it
references were put in the language to make operator overload work right.

- Is there anything that a reference can do that a pointer cannot ???

class MyClass { };

MyClass operator+( const MyClass& lhs, const MyClass& rhs );

Without references, your only choices are to pass by value (which could
be quite expensive for objects of a big class, or pass by pointer which
would make the calling code look like:

MyClass a, b;
MyClass c = &a + &b;

Which seems rather clumsy.

[Phlip]

sarathy wrote:

I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

Normal pointers have too many abilities, so they are high risk. We need a
feature with fewer abilities, so it's safer.

- Is there anything that a reference can do that a pointer cannot ???

It can refer to a temporary object.

It can create a syntax error when you abuse it in ways a pointer would
accept.

Think of a reference as another name for a target - an alias for a target.
Don't think of it as a different kind of pointer.

And, the next time you program a "handle" of some kind, if you don't need it
to be NULL, and don't need to index or increment it, use a reference. Put
another way, always use a reference unless you need a pointers' extra
features.

--
Phlip
http://c2.com/cgi/wiki?ZeekLand <-- NOT a blog!!!

[Roland Pibinger]

On Sat, 29 Jul 2006 20:07:19 GMT, "Daniel T." <da******@earth link.net>
wrote:

>I read "The Design and Evolution of C++" by Bjarne Stroustrup a long
time ago so I may not remember this correctly, however as I remember it
references were put in the language to make operator overload work right.

Interesting. And I thought they were introduced to make C++
programming more convenient and safer. "The Design and Evolution of
C++" really should be my next (and probably last) C++ book.

Best regards,
Roland Pibinger

[Jerry Coffin]

In article <44************ **@news.utanet. at>, rp*****@yahoo.c om
says...

[ ... references ]

Interesting. And I thought they were introduced to make C++
programming more convenient and safer. "The Design and Evolution of
C++" really should be my next (and probably last) C++ book.

Yup, his memory was dead on in this case. D&E, 3.7, says: "References
were introduced primarily to support operator overloading."

--
Later,
Jerry.

The universe is a figment of its own imagination.

[Lally]

Phlip wrote:

sarathy wrote:

I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

By accepting a Foo& instead of Foo*, it's a nice way to get polymorphic
access to objects without having to constantly test for NULL.

[Pierre Barbier de Reuille]

Kai-Uwe Bux wrote:

sarathy wrote:

>Hi,
I have been using C++ for a while. I am not entirely clear with
the concepts of reference in C++.

- Why was there a need for introducing a concept called "Reference" in
C++ when everything was working fine with normal pointers?

It wasn't fine, was it? Pointers *are* tricky. Now, in C++ the trickyness of
pointers is compounded by exceptions: since exceptions can divert the flow
of control at almost any point and toward unknown locations, the basic
requirement of a pointer that every new() is matched by a delete() along
each path of execution is harder to match. Thus, a device was created to
eliminate some uses of pointers. References and standard containers are in
this category. Other devices, like auto_ptr, were introduced to mitigate
the dangers for the remaining cases of pointer use.

>- Is there anything that a reference can do that a pointer cannot ???

References can extend the life-time of temporaries:

But what is the difference between :
[...]
{
log const & ref = create_tmp();
ref.access();
}
[...]

and :

[...]
{
log val = create_tmp();
val.access();
}
[...]

Besides the fact in the first case you won't be able to modify the
reference ?

Pierre

[Alf P. Steinbach]

* Pierre Barbier de Reuille:

Kai-Uwe Bux wrote:

>References can extend the life-time of temporaries:

But what is the difference between :

[...]
{
log const & ref = create_tmp();
ref.access();
}
[...]

and :

[...]
{
log val = create_tmp();
val.access();
}
[...]

Besides the fact in the first case you won't be able to modify the
reference ?

As Kai-Uwe wrote, the reference extends the lifetime of the temporary.
Here the type of the temporary can be a class derived from 'log'. The
non-reference copies the temporary to a variable of type 'log', possibly
slicing.

--
A: Because it messes up the order in which people normally read text.
Q: Why is it such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?