This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: std::map and std::set based on AVL, not RB trees.
Gabriel Dos Reis wrote:
>The way we used to work is not that if one is maintainer one is free
>to implement whatever. We use to work by consensus.
>
>
Then, you don't have my consensus. Because, if we implement that
behavior we have additional complexity (not abstract complexity) in the
code dealing with insertions, i.e., two comparisons (before and after)
instead of one. Also, people following the standard in force may rightly
assume that only one specific comparison is carried out before the full
blown search. I agree that the proposed resolution of DR 322 makes
sense, but, after *5* years is not making progress and only a small
fraction of the users pays attention to the resolutions of the commitee,
most simply expect the behavior mandated in the standard or explained in
widespread books such as Josuttis or Musser et al. Really, I don't see
the point of implementing *now* that behavior, but, again, if you want
to implement it, ok with me. Please, please, move on.
Paolo.
- References:
- std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.
- Re: std::map and std::set based on AVL, not RB trees.