On Sat, Apr 28, 2007 at 08:56:28AM +0100, Richard Jones wrote:
> On Sat, Apr 28, 2007 at 01:12:20AM +0200, Oliver Bandel wrote:
> > On Fri, Apr 27, 2007 at 05:29:11PM +0100, Richard Jones wrote:
> > > The source for COCANWIKI[1] contains extensive support for threading
> > > of mail messages, based on JWZ's algorithm:
> > > 
> > > http://www.jwz.org/doc/threading.html
> > 
> > Nice... you speak of an optimized algorithm for threading.
> > I didn't explored your solution nor did I explored your
> > paper in detail (tomorrow I think I have the time to do it),
> 
> I should point out that the algorithm is due to esteemed hacker[1]
> Jamie Zawinski (http://jwz.org) who used it in Netscape versions 1
> through 3.  They got C++/OO group-think disease from Netscape 4 and
> above (the rot continues to this day).

When I think about netscape, I think about the terrible way,
how it handles newsgroups....
...when doing a resort of the newsgroup-entries it's extremely slow.

Maybe it again checks data on the newsserver again and again.

And when I think about Netscape, I se that it reloads data via network,
when you want to save it, even if it' already on the screen.

So, theese things are the reason, why I don't see why this algorithm
is something special, until I have seen that it really is fast. ;-)

I have thought a while about the threading, and thought I have some good ideas.
A while later, I saw, there already is a datastructure that is useful
for threading, and it is the Tries-datastructure:
  http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Tree/Trie/

If you use the message ID's instead of the char's of a word,
then this datastructure is well suited for news-/mail-threading, I think.


So, other people already had done, what can be used here.

But maybe thze algorithm you show on your pages is really good.

But I would trust your algorithms more than algorithms of
netscape-programmers ;-)
But maybe it's really good and only the overall-design of Netscape was bad. ;-)

Ciao,
   Oliver