Prev: Old fashion function definition?
Next: Design question
From: Andrew Falanga on 18 Dec 2007 11:01
Hi,

I've got a program which with a class structure similar to the
following:

class A
class B : public A
class C : public B

Now, there is a function in class A that must make use of a couple of
functions defined in class B. Before I look at redesigning this code
(because this may be poor design), I'd like to know how one can pass a
reference to a derived class.

Basically, class C calls the public methods of class A and one of
these methods (in A) needs a reference to an object of type B. I
tried simply doing this:

PublicMethodInC( somevalue, *(this) );

but obviously this didn't produce what I needed. I then tried
something like this:

PublicMethodInC( somevalue, *(this->classB) );

which the compiler totally didn't like. Actually, I take that back.
I think it was the linker that complained. (I just checked the output
one more time. I'm still getting used to this M$ stuff, to all the M$
fans in this group, forgive me, but gag me.)

As I think of it though, I could just send in a reference to the class
C object since the B object is derived. From there, the compiler and
linker should link me to the functions I need in B.

I'd still appreciate some suggestions. I hate to admit to it, but I
think I'm going to be redesigning some things. It does seem like poor
design.

Andy
From: Alf P. Steinbach on 18 Dec 2007 11:11
* Andrew Falanga:
>
> I've got a program which with a class structure similar to

Please see the FAQ item on how to post a question about code that
doesn't work, the repost.

Cheers, & hth.,

- Alf


--
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?
From: Jim Langston on 18 Dec 2007 12:23
Andrew Falanga wrote:
> Hi,
>
> I've got a program which with a class structure similar to the
> following:
>
> class A
> class B : public A
> class C : public B
>
> Now, there is a function in class A that must make use of a couple of
> functions defined in class B. Before I look at redesigning this code
> (because this may be poor design), I'd like to know how one can pass a
> reference to a derived class.

Okay, this is a problem. When class A is defined it doesn't know about
class B because it's derieved, doesn't exist yet. If class A needs to call
a method in class B, then class B has to be defined first, but class B needs
to have the definition of class A. You have a recursion problem. What you
should probably do, then, is have the method in class A that needs to know
about class B, put it in class B. Since you have to have class B defined to
call it, it belongs in class B. It most likely needs to be a pure virtual
class. Need to see what it is you're trying to do though.

> Basically, class C calls the public methods of class A and one of
> these methods (in A) needs a reference to an object of type B. I
> tried simply doing this:

So have class C call the public methods of class A, and the one that needs a
reference to class B, put it into class B.

> PublicMethodInC( somevalue, *(this) );
>
> but obviously this didn't produce what I needed. I then tried
> something like this:
>
> PublicMethodInC( somevalue, *(this->classB) );
>
> which the compiler totally didn't like. Actually, I take that back.
> I think it was the linker that complained. (I just checked the output
> one more time. I'm still getting used to this M$ stuff, to all the M$
> fans in this group, forgive me, but gag me.)
>
> As I think of it though, I could just send in a reference to the class
> C object since the B object is derived. From there, the compiler and
> linker should link me to the functions I need in B.
>
> I'd still appreciate some suggestions. I hate to admit to it, but I
> think I'm going to be redesigning some things. It does seem like poor
> design.


Need a little more information on what you are trying to do, but it sounds
like whatever that call is in class A you need to call from C, make pure
virtual in A, define it in B. Class A can not get instancized, but class B
or C can.

--
Jim Langston
tazmaster(a)rocketmail.com


From: Andrew Falanga on 18 Dec 2007 12:27
On Dec 18, 9:11 am, "Alf P. Steinbach" <al...(a)start.no> wrote:
> * Andrew Falanga:
>
>
>
> > I've got a program which with a class structure similar to
>
> Please see the FAQ item on how to post a question about code that
> doesn't work, the repost.
>
> Cheers, & hth.,
>
> - Alf
>

I kind of thought you were asking for code. Next time, please
identify which FAQ you expect people to read. I couldn't find a
question in the FAQ for this group (the only link I have for it is
http://ma.rtij.nl/acllc-c++.FAQ.html), I had to go to the FAQ for
comp.lang.c++ (http://www.parashift.com/c++-faq-lite/). Did this
group, at some time, stop maintaining their own FAQ since the C++ FAQ-
Lite is pretty well maintained?

Because of where I work, I cannot post the code here. So, here goes
in a skeleton program:

#include <iostream>
#include <windows.h>
#include <winsock2.h>
#include <ws2tcpip.h>
#include <cstdlib>

// forward declaration of class B
class B;

class A {
SOCKET sd;
sockaddr sa;
int saLen = sizeof(sockaddr);

public:
void OpenConnection();
int TransferData( B& bObj, char* s , int len ) {
int dataSent;
if( (dataSent = send( sd, s, len )) < 0 ) {
std::cerr << "error in send" << std::endl;
bObj.IncrErrors();
}
return dataSent;
}
};

class B : public A {
static int totalErrorCount;
int myErrorCount;

public:
void IncrErrors( int incrBy = 1 ) {
totalErrorCount = myErrorCount += incrBy;
}
};

class C : public B {
char* str;
int length;
public:
C() : str( "Transfer this string to somewhere" )
{
length = strlen( str );
}

void TestConnection( ) {
OpenConnection();
TransferData( *(this), str, length );
}
};

I know that *(this) refers to this object which in the case I'm using
it refers to the object C, but I was thinking that because it's
derived from object B that it would work. At any rate, I'm still
thinking that some redesign is in order because it seems to me that
it's poor design that leads to objects of type A having to do work
that B should do.

Andy
From: Andrew Falanga on 18 Dec 2007 12:51
On Dec 18, 9:11 am, "Alf P. Steinbach" <al...(a)start.no> wrote:
> * Andrew Falanga:
>
>
>
> > I've got a program which with a class structure similar to
>
> Please see the FAQ item on how to post a question about code that
> doesn't work, the repost.


Well, as happens to me many times, I must now eat a little bit of
crow. I was misinterpreting the linker error (which was an unresolved
symbol). The unresolved symbol that the work "this" in it and from
that I was deducing that my calls to this, as referenced in my last
posting, were causing the problem. However, what was causing the
problem was that there was one function I had somehow managed to skip
over defining. So, in fact, there was an unresolved symbol because
the symbol didn't even exist (except as a declaration in the class
definition). It hurts to be humbled.

Andy
 |  Next  |  Last
Pages: 1 2
Prev: Old fashion function definition?
Next: Design question