[pacman-dev] alpm_list_t

[Nagy Gabor]

Idézés Aaron Griffin <aaronmgriffin at gmail.com>:

> On Nov 21, 2007 9:28 AM, Justin Lampley <jrlampley at gmail.com> wrote:
> > One possible solution is to create a alpm_list_t struct and a
> > alpm_list_node_t struct like Nagy suggested in another thread.  The
> > alpm_list_t struct would contain:
> >
> > struct alpm_list_t {
> > alpm_list_node_t *first;
> > alpm_list_node_t *last;
> > int size; (possibly)
> > }
> 
> Yeah. Dan suggested the same thing at one point. It's fine by me if
> someone wants to do this.
> 
> Here's another suggestion though, to make all our cleanup headaches...
> well, less painful.
> 
> struct alpm_list_t {
>     alpm_list_node_t *first;
>     alpm_list_node_t *last;
>     alpm_list_fn_free *freefn;
>     size_t size;
> }
> 
> then we rework alpm_list_free to do:
> 
> if(list->freefn) {
>     list->freefn(node->data);
> }
> free(node):
> 
> Here's what we gain:
> * The ability to say "free all lists" in all alpm calls, while simply
> setting the freefn to NULL when returning package lists.
> * Less book keeping
> * No need for list freeing macros and knowing when and when not to
> call free_inner
> * No need for free_inner
> 
> Potential con:
> * Requires homogeneous lists that is not enforced in anyway. We do
> this now, but it could introduce more potential error based on the
> fact that we're making this op easier.
> 
> Thoughts?

Other con(?):
This is an answer to both Aaron and Justin's idea.
I like the idea of "list-header" (the new alpm_list_t), but alpm_list_mmerge
won't like that (I refer to this possible hackish solution:
http://www.archlinux.org/pipermail/pacman-dev/2007-November/010065.html)
The problem is, that the list-header adds more bookkeeping to the code and we
lose the current sublist-flexibility. And generating new list-headers to
sublists would lead to many confusion imho.

Bye


----------------------------------------------------
SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu
This mail sent through IMP: http://horde.org/imp/