On Tue, Jun 16, 2009 at 12:48 AM, Thomas Martitz<[email protected]> wrote: > We have the specifications of the original algorithm ( > http://sourcefrog.net/projects/natsort/ ). And that's exactly what the code > in SVN works like currently.
While I would call that rather a description than a specification I do remember that the behaviour of the sorting function was changed several times. I can't remember if it finally got to the same state it had before those changes. However, as it _was_ changed, I need to assume that it's different by now. I can also remember the discussion about breaking the tie if two strings become identical after munging, which is not handled in the original code (there is even a comment about this in the original code). So if the original code doesn't handle this why should the description hold it? Furthermore, have we agreed on that description at all? - Dominik
