Please scroll down to the AlphanumericComparator implementation and observe
that the source is credited to me and has nothing in common with Dave
Koelle's implementation. Jerome, it might be good to alter the comment at
the top of the file to clarify that we use an independent implementation, to
avoid further confusion on this point. I think the change log has the
correct language, crediting Dave for the advocacy but not the code.
Dave's Web site has been a popular destination advocating the usability
advantages of alphanumeric sorting. When this Web site was brought to the
Restlet project's attention, I originally approached Dave Koelle in hopes of
using his existing implementation. However, he was unwilling to release his
work under a Restlet-compatible license (either public domain or CDDL), and
so we could not use it.
Although Dave independently arrived at the idea of an alphanumeric
("natural") sort, he certainly isn't the first to have the idea; the concept
and various implementations appear in computing literature back to at least
the 1970s.
Researching the prior discussion and work on the point led me to choose a
simple character comparison approach with a 2-3x performance gain and memory
footprint advantages over the block comparison approach Dave uses. Dave's
blockwise approach theoretically allows more complex properties to be
examined (for example, comparing hex strings or floats) but these advantages
are not currently exposed in his reference implementation, and necessarily
come with a performance tradeoff.
- Rob
On Mon, May 12, 2008 at 1:20 AM, Leshek <[EMAIL PROTECTED]> wrote:
> Am I looking at a wrong source? Around line 56 of Ditectory.java I see:
>
> * ... The default sorting uses the friendly <a
> *
> href="http://www.davekoelle.com/alphanum.html";>Alphanum<http://www.davekoelle.com/alphanum.html%22%3EAlphanum>Algorithm
> from David
> * Koelle</a>.
>
