Thread: Linked list function

Code:

void traverse (link h, void (*visit)(link)) {
  if (h == NULL) {
    (*visit)(h);
    traverse(h->next, visit);
  }
}

What exactly does that second parameter mean? I have never seen such an argument before.

A pointer to a function.

It might look like:

Code:

void myfunction(link x)
{
    return;
}

/* ... */

/* call traverse */ 
traverse(h, &myfunction);

/* ... */

See http://www.newty.de/fpt/index.html

I get your example, because i have used functions that way before.

But I don't get what (*visit)(h) is. Does it mean the function visit is pointing to the h node

One can also write:

Code:

traverse(h, myfunction);

EDIT:

But I don't get what (*visit)(h) is. Does it mean the function visit is pointing to the h node

visit is a function pointer, so it is just calling the function and passing h as the argument. You can also write visit(h).

Calls what's pointed to by visit with the parameter h.

No need to dereference, it's the same as

Code:

visit(h);

Basically,

Code:

traverse(h, myfunction);        /* short way */
traverse(h, &myfunction);       /* long way */

Code:

visit(h);                       /* short way */
(*visit)(h);                    /* long way */

Both are correct.

Thanks guys.

My lecturer seems intent on confusing people to no end, but then again perhaps it's a good thing I learnt this style of coding. Pretty easy then.

Thanks zacs7 and laserlight =)

Writing the form

Code:

(*visit)(h);

makes sure that it's CLEAR that visit is a funciton pointer - it's not needed, and I don't use that form when I use function pointers - but it is valid, and offers a slightly higher degree of "explaining what is going on" (assuming of course you can follow it).

--
Mats

makes sure that it's CLEAR that visit is a funciton pointer - it's not needed, and I don't use that form when I use function pointers - but it is valid, and offers a slightly higher degree of "explaining what is going on" (assuming of course you can follow it).

The trade off of this slightly higher degree of explanation is that it delves into the details. I think that taking a typedef and then using "normal" function call syntax makes for a higher level of abstraction instead, e.g.,

Code:

typedef void (*VisitFunction)(link);

void traverse(link h, VisitFunction visit) {
  if (h == NULL) {
    visit(h);
    traverse(h->next, visit);
  }
}

Incidentally, h == NULL looks like it should be h != NULL instead since h->next would result in the dereferencing of a null pointer if h == NULL.