[PATCH] Modify 0.7.5beta to do Robinson calculations
David Relson
relson at osagesoftware.com
Sun Oct 20 20:02:17 CEST 2002
At 01:53 PM 10/20/02, Graham Wilson wrote:
>On Sat, Oct 19, 2002 at 02:58:35PM -0400, Greg Louis wrote:
> > On 20021019 (Sat) at 1256:49 -0400, David Relson wrote:
> > > My earlier tests of your Robinson implementation looked like it's a
> > > winner. I think what we should do at the moment is add it and have
> > > an option to use it or the current (original) algorithm. I'm
> > > thinking of '-o' for original algorithm and '-r' for the Robinson
> > > algorithm. How does this sound?
> >
> > Good, to me. I was actually going to patch it that way but then
> > decided that could be a second step, to be done after I have the code
> > better tested. (You might consider -g for Graham and -r for
> > Robinson?)
>
>that sounds more logical than -o and -r. which will be the default?
Graham,
I'll go with "logical" and make the switches '-g' and '-r'. The Graham
algorithm will remain the default for a while. Greg Louis and I are
working to resolve some issues with the Robinson algorithms use of #defines
for values that depend on wordlist sizes (hence need to vary from site to
site).
David
More information about the bogofilter-dev
mailing list