How to decouple a pointer conditional check on link lists?

Hello All,

See the code below. What i'm trying to do is disassociate the condition, so i can re-use the function for different parameters.

Code:

---

// 1. How can I decouple the fast forward condition?
// 2. How can i change the operator?
// 3. What do I do when type of stop condition for structure comparison is no longer int?
// -------------- Attempt 1
// function
void elem_forward_until(seg_list_elem_t ptr, &ptr->s.t_end##, int stop_condition) {
    while (ptr && (ptr->s.t_end## < stop_condition)) {
        ptr = ptr->next;
    }
}
// instance of call
elem_forward_until(ptr, &ptr->s.t_end##, stop_condition);
// -------------- Attempt 2
// function
void elem_forward_until2(seg_list_elem_t ptr, int offset, int stop_condition) {
    while (ptr && ((ptr + offset) < stop_condition)) {
        ptr = ptr->next;
    }
}
// instance of call
elem_forward_until2(ptr, offsetof(struct seg_list_elem_t), stop_condition);

---

Regards,
Wes

Looks like a little non-C (or C++) syntax there.
Maybe you mean something along these lines.

Code:

---

#include <stdio.h>
#include <stdlib.h>
 
typedef struct Segment
{
    int n;
    struct Segment *next;
} Segment;
 
Segment *new_segment(int n, Segment *next)
{
    Segment *seg = malloc(sizeof *seg);
    if (!seg) { perror("new_segment"); exit(EXIT_FAILURE); }
    seg->next = next;
    seg->n = n;
    return seg;
}
 
Segment *prepend(Segment *root, int n)
{
    return new_segment(n, root);
}
 
Segment *append(Segment *root, int n)
{
    Segment *seg = new_segment(n, NULL);
    if (!root) return seg;
    Segment *s = root;
    while (s->next) s = s->next;
    s->next = seg;
    return root;
}
 
void print(const Segment *segs)
{
    for ( ; segs; segs = segs->next) printf("%d ", segs->n);
    printf("\n");
}
 
int cmp_less  (const Segment *a, const void *b) { return a->n < *(int*)b; }
int cmp_greater(const Segment *a, const void *b) { return a->n > *(int*)b; }
 
Segment *elem_forward_until(
    Segment*    segs,
    int(*      cmp)(const Segment*, const void*),
    const void* stop)
{
    while (segs && cmp(segs, stop)) segs = segs->next;
    return segs;
}
 
int main()
{
    const int init[] = {4, 5, 3, 6, 2, 7, 1, 9, -1}; // -1 marks end of init list
 
    Segment *segs = NULL;
    for (const int *n = init; *n != -1; ++n) segs = append(segs, *n);
    print(segs);
 
    Segment *s = elem_forward_until(segs, cmp_less, &(int){7});
    if (s) printf("%d\n", s->n); else printf("nil\n");
 
    s = elem_forward_until(segs, cmp_greater, &(int){1});
    if (s) printf("%d\n", s->n); else printf("nil\n");
 
    return 0;
}

---

Just a note: I prefer to deal with lists as described by Sedgewick because we can create polymorphic nodes, in C. Here's an example using a circular single linked list:

Code:

---

#include <stdio.h>
#include <stdlib.h>

// A list always have a 'head' sentinel node. All data nodes are polymorphic since
// we only need to declare a structure with the same 'next' pointer at the beginning.

// Used to initialize a list
#define SLIST_HEAD_PTR struct slist_head *next

struct slist_head { struct slist_head *next; };

#define SLIST_INIT_HEAD(head_) { &(head_) }

static inline void slist_add_next( struct slist_head *node, struct slist_head *new )
{
  new->next = node->next;
  node->next = new;
}

static inline struct slist_head *slist_del_next( struct slist_head *node )
{
  struct slist_head *next = node->next;
  node->next = next->next;
  return next;
}

static inline int slist_isempty( struct slist_head *head )
{
  return head->next == head;
}

#define slist_foreach( iter_, head_ ) \
  for ( iter_ = (head_)->next; iter_ != head_; iter_ = (iter_)->next )

// ------- test ------------

int main( void )
{
  // My node (polymorphism)
  struct mynode {
    SLIST_HEAD_PTR;    // it is important this is the first declaration!
    int x;
  }; 

  struct slist_head list = SLIST_INIT_HEAD( list );  // our head of the list.
  struct slist_head *p;
  struct mynode *q;

  // Notice: Allocation isn't present at the list routines!
  q = malloc( sizeof *q );
  q->x = 1;
  slist_add_next( &list, (struct slist_head *)q );  // q is tail.
  p = (struct slist_head *)q;                      // save the tail node.

  q = malloc( sizeof *q );
  q->x = 2;
  slist_add_next( p, (struct slist_head *)q );      // q is tail.
  p = (struct slist_head *)q;                      // save the tail node.

  q = malloc( sizeof *q );
  q->x = 3;
  slist_add_next( p, (struct slist_head *)q );      // q is tail.

  // List all nodes.
  slist_foreach( p, &list )
    printf( "%d\n", ((struct mynode *)p)->x );

  // delete all nodes.
  // Notice: Deallocation isn't present at list routines either.
  while ( ! slist_isempty( &list ) )
    free( slist_del_next( &list ) );
}

---

Who is the this "Sedgewich"? I really think they should move off to other languages..

Is it really so hard to recognize or type this pattern?

Code:

---

  struct whatever *p = first_p;
  while(p != NULL) {
      // dp spmething
      p = p->next;
  }

---

Wait till they need a structure to be a member of two different lists at the same time...

Code:

---

struct statistic_year {
  struct x *next_for_year;
  struct x *next_for_statistic;
  double value;
};.

---

Have a look at this mid-80s magazine, https://vintageapple.org/byte/pdf/19...ss_Storage.pdf on page 137 (by the page number) if you want to see where this ends...

Code:

---

In C:
if ((c > = 'a' && c < = 'z') II (c > = 'A' && c < = 'Z'))
  return(LETTER);

In Easy C:

if ((c GE 'a' AND c LE 'z) OR (c GE 'A' AND c LE 'Z'))
  return(LETTER);

---

And these gems:

Code:

---

#define CASE(e) { switch (e) { /* head of case /*
#define CASEOF(e) casee : { /* case block • /
#define DEFCASE default : { /* default case block • /
#define ENDCOF } break ; /* end of case block • /
#define ENDCASE } } 1· end of case ·1

---

IMO excessively leaning on the pre-processor and inline functions rather than changing languages is a bad idea.

Can someone break this down barney style for me - "*(int*)b"

Hamster_nz? Are you the same guy whos posted lots of FPGA related topics?

First question....

cast 'b' to be a pointer to int, then dereference it to return the integer at memory address b.

Code:

---

int a = *((int *)b);

---

Second qn, yes, that was me a decade ago. Stopped now because I got sick of being begged or shouted at in emails for not helping students with their assignments.

Regarding a = *((int*)b); Is this so b can be a different data type. For example char?

Is it not constrained by data type underlined.
returna->n < *(int*)b;

Nz, shame to hear. You're a bit of a legend

Yes
,

Code:

---

  char *buffer = malloc(2000); // Allocate a buffer, enough for 2000 chars, or 500 32-bit ints
  // ... do something here to fill the buffer
  int32_t a = *((int32_t *)buffer);  // a will be the  value of the first four bytes of the buffer, treated as an 32-bit signed integer.

---

Doing things like this is somewhat bad form, as dynamically changing pointer types can break some of the finer points of C, esp compiler optimizations which assume that different data types don't overlap each other in memory (that is a very bad explanation of strict aliasing)

A union is far better way, that is legal C.

Thanks all,

How do I mark this as solved?

Regards,
Wes

There's no actual mark as solved.
Saying thanks is usually enough and people stop replying.