![]() |
|
![]() |
|
|
Thread Tools | Display Modes |
|
|
#1 |
|
The Oblivious One
Join Date: May 2005
Location: Ontario, Canada
Posts: 648
Rep Power: 4
![]() |
Appreciate comments on some linked list code
As a beginner, I thought it would be neat to see if I could create some somewhat reusable linked list code. You'll need rake ( http://docs.rubyrake.org/ ) to build it, and I have included some API docs using doxygen.
This was mostly for fun, so I haven't done any real testing. However, of the tests I have run, valgrind reports 0 memory leaks and 0 errors. I am posting this to get some feedback on my coding style, and maybe learn of some best practises in C I don't yet know. I would appreciate any advice you could provide, though I'm not expecting anybody to take this too seriously. Make changes to main.c and simply run rake to alter the program. On a side note, how would I make this code easily include-able in other programs? I realize that my int-dependant version is not that useful, but I would like to know for future application. Regardless, thanks for looking. ![]()
__________________
Dr. Zoidberg: [ecstatic] I'm going to a movie... with FRIENDS! |
|
|
|
|
|
#2 |
|
Professional Programmer
![]() Join Date: Sep 2005
Posts: 419
Rep Power: 4
![]() |
At a glance, it looks pretty good. However, there are a few points that I would like to make.
![]() 1) At a minor space cost in the list object, you can save the length of the list and not have to recalculate it. That turns list_length from an O(n) operation to an O(1) operation. 2) Operations that are strictly owned by the library (eg. the new_node function) should be static so that they're only visible to the library. 3) Libraries should be minimal and extensible. That means that you think really hard about your design so that you can give people the smallest possible library that's safe, and then let them extend it for things that aren't absolutely required in the base library. A good example is C++'s iterators. If you provide something like that as a framework, you can make a separate algorithms library that holds a lot of the functionality that you're hardcoding now. That promotes reuse and makes your maintenance job easier. The downside is that you have to work harder on the design. ![]() Something like this might be interesting: #include <stdlib.h>
typedef struct jsw_node *jsw_list_iterator;
// Private helpers for the library
static void *move_next ( void *obj );
static void *get_data ( void *obj );
static void *set_data ( void *obj, void *data );
static void *add_node ( void *obj, void *data );
static struct jsw_node *new_node ( void *init_data, struct jsw_node *init_next );
struct jsw_node {
void *_data;
struct jsw_node *_next;
// Forward iterator behaviors
void *(*next) ( void *obj );
void *(*data) ( void *obj );
void *(*set) ( void *obj, void *data );
void *(*add) ( void *obj, void *data );
};
struct jsw_list {
struct jsw_node *head;
};
static void *move_next ( void *obj )
{
return ((struct jsw_node *)obj)->_next;
}
static void *get_data ( void *obj )
{
return ((struct jsw_node *)obj)->_data;
}
static void *set_data ( void *obj, void *data )
{
struct jsw_node *node = obj;
void *old_data = node->_data;
node->_data = data;
return old_data;
}
static void *add_node ( void *obj, void *data )
{
struct jsw_node *node = obj;
struct jsw_node *nn = new_node ( data, node->_next );
node->_next = nn;
return nn;
}
static struct jsw_node *new_node ( void *init_data, struct jsw_node *init_next )
{
struct jsw_node *node = malloc ( sizeof *node );
if ( node == NULL )
return NULL;
node->_data = init_data;
node->_next = init_next;
node->next = move_next;
node->data = get_data;
node->set = set_data;
node->add = add_node;
return node;
}
struct jsw_list *new_list ( void )
{
struct jsw_list *list = malloc ( sizeof *list );
if ( list == NULL )
return NULL;
list->head = new_node ( NULL, NULL );
if ( list->head == NULL ) {
free ( list );
return NULL;
}
return list;
}
jsw_list_iterator jsw_list_begin ( struct jsw_list *list )
{
// Before the first real node
return list->head;
}
jsw_list_iterator jsw_list_end ( struct jsw_list *list )
{
// After the last real node
return NULL;
}
// Quickie test, no cleanup done
#include <stdio.h>
void *make_data ( int x )
{
int *new_data = malloc ( sizeof *new_data );
*new_data = x;
return new_data;
}
int main ( void )
{
struct jsw_list *list = new_list();
if ( list != NULL ) {
jsw_list_iterator it = jsw_list_begin ( list );
jsw_list_iterator end;
int i;
for ( i = 1; i < 7; i++ )
it = it->add ( it, make_data ( i ) );
it = jsw_list_begin ( list );
end = jsw_list_end ( list );
while ( ( it = it->next ( it ) ) != end )
printf ( "%d ", *(int *)it->data ( it ) );
printf ( "\n" );
}
return 0;
}
__________________
Even if the voices aren't real, they have some pretty good ideas. Last edited by Narue; Jul 20th, 2006 at 11:10 AM. |
|
|
|
|
|
#3 |
|
The Oblivious One
Join Date: May 2005
Location: Ontario, Canada
Posts: 648
Rep Power: 4
![]() |
Thanks Narue. Helpful as usual.
![]() A few things I'll be doing in the future:
Thanks again ![]() EDIT: There's a new forum highlighting tag. use [ highlight=c ] and [ /highlight ] which uses syntax highlighting. ![]()
__________________
Dr. Zoidberg: [ecstatic] I'm going to a movie... with FRIENDS! |
|
|
|
|
|
#4 |
|
Professional Programmer
![]() Join Date: Sep 2005
Posts: 419
Rep Power: 4
![]() |
>prefix with something like jhk_<function> to avoid naming clashes.
Always a good idea in the absence of namespaces. >prefix important struct fields with underscores. Be wary of leading underscores. There are times when they're reserved. >use [ highlight=c ] and [ /highlight ] which uses syntax highlighting. Not until it preserves my precise formatting. ![]()
__________________
Even if the voices aren't real, they have some pretty good ideas. |
|
|
|
|
|
#5 |
|
Expert Programmer
Join Date: Jun 2005
Posts: 894
Rep Power: 4
![]() |
I've only looked at the linked.c so far, but the first function caught my eye:
int one_node(const struct LinkedList *list) {
return list->first->next == NULL;
}You should also put some sort of error checking on your calls to malloc. In new_linked_list, you don't need all the if tests, just push nodes for whatever ints are passed in In delete_linked_list, again all 3 paths through the code end up doing the same thing, so you don't need the special cases for 0 and 1 nodes. Also delete_linked_list should set the first pointer to NULL, so you can reuse the list straight away. Actually the same goes for a lot (but not all) of the tests for one_node and empty lists. Similarly to Narue's point, having a "last" or "tail" pointer as well as a "first" pointer in your linked list will make operations like last_node (and hence push_node and everything that use that) a lot faster. It does lead to more housekeeping by your library functions, however. |
|
|
|
|
|
#6 |
|
The Oblivious One
Join Date: May 2005
Location: Ontario, Canada
Posts: 648
Rep Power: 4
![]() |
Thanks, The Dark.
![]() I didn't catch a lot of those. :o
__________________
Dr. Zoidberg: [ecstatic] I'm going to a movie... with FRIENDS! |
|
|
|
![]() |
| Bookmarks |
| Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| dev c++ software, template problem | cairo | C++ | 11 | Jun 2nd, 2006 1:42 PM |
| Singly Linked List Help | Firebar | Java | 3 | May 22nd, 2005 11:56 AM |
| User-defined creatNode and deleteNode functions for a doubly-linked list | jgs | C | 2 | Apr 28th, 2005 9:53 AM |
| Linked List Part 2 | Nish | C++ | 0 | Mar 4th, 2005 4:56 PM |
| airport Log program using 3D linked List : problem reading from file | gemini_shooter | C++ | 0 | Mar 2nd, 2005 5:12 PM |