«

Ok, this may sound confusing....but it's really simple. If you're confused,
just look at my example code and it'll make sense.

Here's my situation. I have 2 classes....A and B.

Class A has a member variable of type B.
Class B has a member function which does calculations which are dependent
upon values of members of the Class A that owns it.

When in that function of class B, I don't know how I can gain access to it's
parent Class A's member variables, so my solution to the problem is to take
as a parameter to that function, a class A datatype. So, it's like this:

class A
{
public:
int x;
B myB;
:
:
};

class B
{
public:
int y;
void theFunction(A parentA);
:
:
};

void B::theFunction(A parentA){
y = parentA.x * 10;
}

I would call it like this:
A myA;
myA.myB(myA);

Now, the problem is, both A.h and B.h have to include eachother. When they
do, I get the following compilation errors:
A.h(10): error C2143: syntax error : missing ';' before 'B'
A.h(10): error C2501: 'A::B' : missing storage-class or type specifiers

Line 10 is the line that I declare B myB is A.h.

If I comment out theFunction in B and I comment out the include of A.h,
it'll compile no problem. But that obviously isn't the solution as B can't
use A.

So, I need one of 2 solutions. Either (and most preferably) I somehow gain
access to the parent object A's member variables, without having to pass A
as a parameter to B's function....or, I keep it like it is, but somehow get
it to compile.

Can anyone inform me how either of the above can be accomplished?

Oh...and I know I could just pass myA.x, but don't want to do that as the
real function is called often and is actually dependent upon many member
variables of A.

Thanks a lot for any help!

"Eternally" <> wrote...
> Ok, this may sound confusing....but it's really simple. If you're
confused,
> just look at my example code and it'll make sense.
>
> Here's my situation. I have 2 classes....A and B.
>
> Class A has a member variable of type B.
> Class B has a member function which does calculations which are dependent
> upon values of members of the Class A that owns it.
>
> When in that function of class B, I don't know how I can gain access to
it's
> parent Class A's member variables, so my solution to the problem is to
take
> as a parameter to that function, a class A datatype.

That's one way to do it...

An alternative way would be to pass the 'A*' during construction
of the member B, and then B will always know where its parent is.

> So, it's like this:
>
> class A
> {
> public:
> int x;
> B myB;
> :
> :
> };
>
> class B
> {
> public:
> int y;
> void theFunction(A parentA);

I wouldn't pass by value, of course. It is better to pass by
reference, for example. All you need to declare is that 'A' is
a class, really. This should help:

void theFunction(class A parentA);

and you don't need to include the declaration of A.

> :
> :
> };
>
> void B::theFunction(A parentA){
> y = parentA.x * 10;
> }
>
> I would call it like this:
> A myA;
> myA.myB(myA);
>
> Now, the problem is, both A.h and B.h have to include eachother. When
they
> do, I get the following compilation errors:
> A.h(10): error C2143: syntax error : missing ';' before 'B'
> A.h(10): error C2501: 'A::B' : missing storage-class or type specifiers
>
> Line 10 is the line that I declare B myB is A.h.
>
> If I comment out theFunction in B and I comment out the include of A.h,
> it'll compile no problem. But that obviously isn't the solution as B
can't
> use A.
>
> So, I need one of 2 solutions. Either (and most preferably) I somehow
gain
> access to the parent object A's member variables, without having to pass A
> as a parameter to B's function....or, I keep it like it is, but somehow
get
> it to compile.

See above. The alternative solution I was talking about is to
declare 'B's constructor as accepting A&:

class A; // forward declataion
class B {
A& parent;
public:
B(A& p) : parent(p);
};

....
class A {
B myB;
public:
A() : myB(*this) {}
...
};

The compiler may not line the use of 'this' in the initialiser list,
but you can ignore the warning.

> Can anyone inform me how either of the above can be accomplished?
>
> Oh...and I know I could just pass myA.x, but don't want to do that as the
> real function is called often and is actually dependent upon many member
> variables of A.

Shouldn't it be a member of A, then?

Victor

"Victor Bazarov" <> wrote in message
news:...
> "Eternally" <> wrote...
> > Ok, this may sound confusing....but it's really simple. If you're
> confused,
> > just look at my example code and it'll make sense.
> >
> > Here's my situation. I have 2 classes....A and B.
> >
> > Class A has a member variable of type B.
> > Class B has a member function which does calculations which are
dependent
> > upon values of members of the Class A that owns it.
> >
> > When in that function of class B, I don't know how I can gain access to
> it's
> > parent Class A's member variables, so my solution to the problem is to
> take
> > as a parameter to that function, a class A datatype.
>
> That's one way to do it...
>
> An alternative way would be to pass the 'A*' during construction
> of the member B, and then B will always know where its parent is.
>
> > So, it's like this:
> >
> > class A
> > {
> > public:
> > int x;
> > B myB;
> > :
> > :
> > };
> >
> > class B
> > {
> > public:
> > int y;
> > void theFunction(A parentA);
>
> I wouldn't pass by value, of course. It is better to pass by
> reference, for example. All you need to declare is that 'A' is
> a class, really. This should help:
>
> void theFunction(class A parentA);
>
> and you don't need to include the declaration of A.
>
> > :
> > :
> > };
> >
> > void B::theFunction(A parentA){
> > y = parentA.x * 10;
> > }
> >
> > I would call it like this:
> > A myA;
> > myA.myB(myA);
> >
> > Now, the problem is, both A.h and B.h have to include eachother. When
> they
> > do, I get the following compilation errors:
> > A.h(10): error C2143: syntax error : missing ';' before 'B'
> > A.h(10): error C2501: 'A::B' : missing storage-class or type specifiers
> >
> > Line 10 is the line that I declare B myB is A.h.
> >
> > If I comment out theFunction in B and I comment out the include of A.h,
> > it'll compile no problem. But that obviously isn't the solution as B
> can't
> > use A.
> >
> > So, I need one of 2 solutions. Either (and most preferably) I somehow
> gain
> > access to the parent object A's member variables, without having to pass
A
> > as a parameter to B's function....or, I keep it like it is, but somehow
> get
> > it to compile.
>
> See above. The alternative solution I was talking about is to
> declare 'B's constructor as accepting A&:
>
> class A; // forward declataion
> class B {
> A& parent;
> public:
> B(A& p) : parent(p);
> };
>
> ...
> class A {
> B myB;
> public:
> A() : myB(*this) {}
> ...
> };
>
> The compiler may not line the use of 'this' in the initialiser list,
> but you can ignore the warning.
>
> > Can anyone inform me how either of the above can be accomplished?
> >
> > Oh...and I know I could just pass myA.x, but don't want to do that as
the
> > real function is called often and is actually dependent upon many member
> > variables of A.
>
> Shouldn't it be a member of A, then?
>
> Victor
>
>

Hi,

That's a nice solution, and I'll probably use it, but the problem still is
that even if I comment out theFunction in B and all references to A, as long
as B is including A.h, those compiler errors still remain. If I comment out
#include "A.h", then it compiles without errors....but if I put it in the
errors come back.

If A.h includes B.h and B.h includes A.h, then those errors will be there.

Thanks for the help!

"Eternally" <> wrote...
> [...]
> That's a nice solution, and I'll probably use it, but the problem still is
> that even if I comment out theFunction in B and all references to A, as
long
> as B is including A.h, those compiler errors still remain. If I comment
out
> #include "A.h", then it compiles without errors....but if I put it in the
> errors come back.

"Doctor, if I do this, it hurts. What should I do?"
"Don't do that."

> If A.h includes B.h and B.h includes A.h, then those errors will be there.

Of course. The solution is not to have those circular includes.

Victor

"Eternally" <> wrote in message
news:k7FPa.33262$...
> Ok, this may sound confusing....but it's really simple. If you're
confused,
> just look at my example code and it'll make sense.
>
> Here's my situation. I have 2 classes....A and B.
>
> Class A has a member variable of type B.
> Class B has a member function which does calculations which are dependent
> upon values of members of the Class A that owns it.
>
> When in that function of class B, I don't know how I can gain access to
it's
> parent Class A's member variables, so my solution to the problem is to
take
> as a parameter to that function, a class A datatype. So, it's like this:
>
> class A
> {
> public:
> int x;
> B myB;
> :
> :
> };
>
> class B
> {
> public:
> int y;
> void theFunction(A parentA);
> :
> :
> };
>
> void B::theFunction(A parentA){
> y = parentA.x * 10;
> }
>
> I would call it like this:
> A myA;
> myA.myB(myA);
>
> Now, the problem is, both A.h and B.h have to include eachother. When
they
> do, I get the following compilation errors:
> A.h(10): error C2143: syntax error : missing ';' before 'B'
> A.h(10): error C2501: 'A::B' : missing storage-class or type specifiers
>
> Line 10 is the line that I declare B myB is A.h.
>
> If I comment out theFunction in B and I comment out the include of A.h,
> it'll compile no problem. But that obviously isn't the solution as B
can't
> use A.
>
> So, I need one of 2 solutions. Either (and most preferably) I somehow
gain
> access to the parent object A's member variables, without having to pass A
> as a parameter to B's function....or, I keep it like it is, but somehow
get
> it to compile.
>
> Can anyone inform me how either of the above can be accomplished?
>
> Oh...and I know I could just pass myA.x, but don't want to do that as the
> real function is called often and is actually dependent upon many member
> variables of A.
>
> Thanks a lot for any help!
>
>

Something I have done in the face of circular dependencies is as illustrated
below. Im not especialy proud of it but it worked. If anyone has comments on
the technique Id be glad to hear them.

---- A.h ----

#ifndef A_H
#define A_H

class B; /* #include "B.h" */

class A
{ private:
B myB;
....
};
.....

/**
* By this point itll be possible to define the
* B class since A has already been defined
*/
#include "B.h"

#endif

---- B.h ----

#ifndef B_H
#define B_H

#include "A.h"

class B
{ public:
void theFunction(A*);
....
};
.....

#endif
----

"Ellarco" <> wrote in message
news:1pRPa.22683$...
>
> "Eternally" <> wrote in message
> news:k7FPa.33262$...
> > Ok, this may sound confusing....but it's really simple.
If you're
> confused,
> > just look at my example code and it'll make sense.
> >
> > Here's my situation. I have 2 classes....A and B.
> >
> > Class A has a member variable of type B.
> > Class B has a member function which does calculations
which are dependent
> > upon values of members of the Class A that owns it.
> >
> > When in that function of class B, I don't know how I can
gain access to
> it's
> > parent Class A's member variables, so my solution to the
problem is to
> take
> > as a parameter to that function, a class A datatype.
So, it's like this:
> >
> > class A
> > {
> > public:
> > int x;
> > B myB;
> > :
> > :
> > };
> >
> > class B
> > {
> > public:
> > int y;
> > void theFunction(A parentA);
> > :
> > :
> > };
> >
> > void B::theFunction(A parentA){
> > y = parentA.x * 10;
> > }
> >
> > I would call it like this:
> > A myA;
> > myA.myB(myA);
> >
> > Now, the problem is, both A.h and B.h have to include
eachother. When
> they
> > do, I get the following compilation errors:
> > A.h(10): error C2143: syntax error : missing ';' before
'B'
> > A.h(10): error C2501: 'A::B' : missing storage-class or
type specifiers
> >
> > Line 10 is the line that I declare B myB is A.h.
> >
> > If I comment out theFunction in B and I comment out the
include of A.h,
> > it'll compile no problem. But that obviously isn't the
solution as B
> can't
> > use A.
> >
> > So, I need one of 2 solutions. Either (and most
preferably) I somehow
> gain
> > access to the parent object A's member variables,
without having to pass A
> > as a parameter to B's function....or, I keep it like it
is, but somehow
> get
> > it to compile.
> >
> > Can anyone inform me how either of the above can be
accomplished?
> >
> > Oh...and I know I could just pass myA.x, but don't want
to do that as the
> > real function is called often and is actually dependent
upon many member
> > variables of A.
> >
> > Thanks a lot for any help!
> >
> >
>
> Something I have done in the face of circular dependencies
is as illustrated
> below. Im not especialy proud of it but it worked. If
anyone has comments on
> the technique Id be glad to hear them.
>
> ---- A.h ----
>
> #ifndef A_H
> #define A_H
>
> class B; /* #include "B.h" */
>
> class A
> { private:
> B myB;
> ....
> };
> ....
>
> /**
> * By this point itll be possible to define the
> * B class since A has already been defined
> */
> #include "B.h"
>
> #endif
>
> ---- B.h ----
>
> #ifndef B_H
> #define B_H
>
> #include "A.h"
>
> class B
> { public:
> void theFunction(A*);
> ....
> };
> ....
>
> #endif
> ----
>

You've got the forward declaration and #include-in-header
backwards.

"TR" <> writes:

> "Victor Bazarov" <> wrote in message
> newsNKPa.37554$ et...
> > > If A.h includes B.h and B.h includes A.h, then those errors will be
> there.
> >
> > Of course. The solution is not to have those circular includes.
>
> A better long-term solution is header guards, unless this is the most
> complex program he's ever going to create.
>
> Use this in your A.h and B.h files:
>
> #ifndef FILENAME_H
> #define FILENAME_H
>
> contents of header
>
> #endif
>
> or in VC++ just:
> #pragma once
>
> at the top. But it's less portable.

Huh? Include guards don't prevent circular includes - have you
actually *READ* the thread?

regards
frank

--
Frank Schmitt
4SC AG phone: +49 89 700763-0
e-mail: frank DOT schmitt AT 4sc DOT com

I haven't read the whole thread -- so I am sorry if the question has
been answered already.

When the class A is declared in a header pointer members of A need not
have been defined (only declared) a this point:

file A.h:

class B; // forward declaration

class A {

B* b;
int f(B x);
}


The forward declaration allows you to use the name "B"
as a type name, A.h need not include B.h.
The class B has to be defined somewhere of course.
You cannot USE B in the header, ie. this won't work

class A {

B* b;
int f(B x){ return b.doSomething(); }

}

But if the header contains only declarations and no implementations
that problem does not occur. You move the definition of A::f to A.cc
and #include B.h in A.cc.