loop in strucures definition

[riccardo]

Hi,
I've

typedef struct {
foo2 a;
} foo1;

typedef struct {
foo1 b;
} foo2;


foo2 is not defined yet when used to define foo1.
Placing the foo2 definition before foo1 doesn't help of course.
How can I solve this in a clean way?

R

 

[Noob]

I've

typedef struct {
foo2 a;
} foo1;

typedef struct {
foo1 b;
} foo2;


foo2 is not defined yet when used to define foo1.
Placing the foo2 definition before foo1 doesn't help of course.
How can I solve this in a clean way?

Can you tell what value bar returns?

int bar(void) { return bar( ); }

 

[Felix Palmen]

Can you tell what value bar returns?

int bar(void) { return bar( ); }

nothing, because it does not terminate. How do you think this relates to
the OP's question?

 

[Noob]

nothing, because it does not terminate. How do you think this relates to
the OP's question?

How big do you think a foo1 is?

 

[John Bode]

Hallo riccardo,




Declare the structs first and use these declarations inside.

struct foo1;
struct foo2;

struct foo1
{
  struct foo2 a;

}

Unfortunately, since "struct foo2" is still incomplete at this stage,
you can't declare an instance of it in "struct foo1"; you *can*
declare a pointer to it, though:

struct foo1
{
struct foo2 *a;
};

Circular dependencies are bad juju and should be avoided if possible.

 

[Felix Palmen]

Unfortunately, since "struct foo2" is still incomplete at this stage,
you can't declare an instance of it in "struct foo1"; you *can*
declare a pointer to it, though:

struct foo1
{
struct foo2 *a;
};

Circular dependencies are bad juju and should be avoided if possible.

Oh right, I completely missed that one. Indeed, trying to "circular
embed" structures is a paradoxon

Regards,
Felix

 

[Felix Palmen]

How big do you think a foo1 is?

Haha, seems I silently added the pointer somewhere in my conscience so
it looked sane to me.

 

[riccardo]

Unfortunately, since "struct foo2" is still incomplete at this stage,
you can't declare an instance of it in "struct foo1"; you *can*
declare a pointer to it, though:

struct foo1
{
struct foo2 *a;
};

Tricky but effective!
didn't know I did step on such a nasty topic..
hope it was not a FAQ bible question..
RM

 

[riccardo]

You have an object, which we will call a "box". This box contains things
within it, including another box. Since the definition of "box" is
something which contains another box, this second box contains a third
box. The third must contain a fourth, which contains a fifth, which
contains a six, and so on.

Now, if instead of the box including another box, it includes a piece of
paper telling you where another box may be found (or the word "NULL"),
then this infinitely-recursive definition never occurs.

Thanks,
Very good example (I must say I have some reason to have the recursion).
RM

 

[lawrence.jones]

typedef struct {
foo2 a;
} foo1;

typedef struct {
foo1 b;
} foo2;

foo2 is not defined yet when used to define foo1.
Placing the foo2 definition before foo1 doesn't help of course.
How can I solve this in a clean way?

Method 1:

typedef struct foo1 {
struct foo2 a;
} foo1;

typedef struct foo2 {
struct foo1 b;
} foo2;

Method 2:

typedef struct foo1 foo1;
typedef struct foo2 foo2;
struct foo1 {
foo2 a;
};

struct foo2 {
foo1 b;
};

 

[Niklas Holsti]

Such a structure would have indefinite size. To make recursive structure, you
need to make at least one reference in the loop into a pointer. (In Algol 68
terminology, the type has to have yin yang.)

Hey, thanks for bringing tears of nostalgia to my eyes! Did you read the
two-level Algol 68 grammar? Did you also, as I did, feel like you were
reading a Prolog interpreter for Algol 68, or at least semantic analyser?

 

[Ben Bacarisse]

Method 1:

typedef struct foo1 {
struct foo2 a;
} foo1;

typedef struct foo2 {
struct foo1 b;
} foo2;

Method 2:

typedef struct foo1 foo1;
typedef struct foo2 foo2;
struct foo1 {
foo2 a;
};

struct foo2 {
foo1 b;
};

I think your were expecting a (and possibly b) to be a pointer. As it
stands, in both your methods struct foo1 has a member with incomplete
type.

 

[Paul N]

Tricky but effective!
didn't know I did step on such a nasty topic..
hope it was not a FAQ bible question..

FAQ 1.15 seems to be the nearest...