When i understood the current pacman code, there is a cycle check, therefore i don't see a reason why ->last should not be provided. Though, i don't completely understand when you build a linear depend list why you should want to step backwards anyway, especially not in Ss mode. On 3/5/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 3/4/07, Jürgen Hötzel <juergen@hoetzel.info> wrote:
due the missing "last element hint" which speeded up the add operation before aaron introduced the public alpm_list type two months ago. I consider adding the "last element" member again. Any objections?
Actually, that's probably a bad idea. It's almost trivial to convert the list to a circular one. Then alpm_list_last(foo) would just be foo->prev;
There should never be a case when this would cause a problem (i.e. some sort of "scan backward until ->prev is null" function would only be useful in the case of bad design). _______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://www.archlinux.org/mailman/listinfo/pacman-dev