man page problems

[David Relson]

On Tue, 09 Dec 2003 13:46:54 +0100
Boris 'pi' Piwinger <3.14 at logic.univie.ac.at> wrote:

> Hi!
> 
> I just sent David a patch for the man page which clarifies
> and updates some points. Some things are not yet clear to
> me, so I'd also like to do those, but I need your help:
> 
> The man page describes how bogofilter started from Paul's
> article, how that was improved, Gary's contribution, further
> improvements to Fisher. The latter has no further
> explanation or link. Can someone give a short description or
> a link for it?

Gary has suggested two algorithm.  We generally call them the "robinson"
and "fisher" algorithms.  As Greg often points out, more accurate names
are the "robinson geometric mean" and the "robinson fisher" algorithms. 
In the second one, Gary added a reverse chi-square calculation to the
end of the computation.  My understanding is that the reverse chi-square
technique was developed by Fisher for other purposes and applied by
Robinson to bayesian calculations.


> For -B it is not clear to me how to use it with multiple
> files. Can I just list those at the end of the command line?
> Anyhow, the SYNOPSIS lists the filename directly with the
> option which is not needed. -M does not list a filename
> there at all. How should this be solved?

When -B is used, all non-switch options are treated as filenames.  Just
throw them in at the end of the command line.

> -F needs further explanation. How can this be used? In which
> context?

Some info, like the histogram triggered by '-vv', is only printed for
messages with high scores.  '-F' will "force" the printing of that info
without regard to the score.  I'd have to look at the code to see
exactly where the flag is used.  

> What are possible warning for -q? I've never seen one.

Interesting.  '-q' is defined for bogofilter and bogolexer.  However
bogofilter never uses it, so I've deleted it.

I also notice that bogotune was using '-Q' for the same purpose.  I've
changed it to '-q'.

> -x should give a hint where those flags can be found.

They're defined in debug.h.  As they're a debug aid for the developers,
the exact meaning of the various flags will change over time as we
encounter a need to display new information or decide we no longer need
to see certain info.

> The AUTHOR section is really not accurate. Maybe the AUTHORS
> file could be included? Or just list original author and
> current maintainer?

Good suggestion.  The AUTHOR should probably be ESR, Matthias, Greg, and
me.  The AUTHORS file is the more comprehensive list, including
contributors.