bogofilter (-o) & bogoutil (-c) bugs?

[David Relson]

At 12:05 PM 2/17/03, W M Brelsford wrote:

>On Mon Feb 17 2003 at 10:53 AM -0500, David Relson wrote:
> > There's _no_ need for keeping 0-count tokens.  They just take up
> > space.  They'll go away.
>
>Good.
>
> > Have you observed any other details that need attending to?
>
>No, that was my list.  But you didn't answer my last question
>regarding -o without -C: don't command-line arguments override
>configuration file settings?

Bill,

I overlooked that one.  Guess I better get my eyeglasses checked.

Since bogofilter doesn't track whether values come from the command line or 
from the config file, it uses the last value it sees.  Thus if spam_cutoff 
is specified in bogo.cf, "bogofilter -c bogo.cf -o .44 -qv" will give 
different results than "bogofilter -o .44 -c bogo.cf -qv".  When the code 
was written, I couldn't think of a good way (read "simple, clean, and 
elegant") to code a precedence like you suggest.  Have you any ideas on 
coding it?

>Has any thought been given to an option to specify additional
>regular expressions for delimiting tokens?  My wife tends to use a
>random number of periods to separate clauses, so replies that quote
>her message leave lots of tokens like "soon.....but".  Not a big
>deal -- I'll probably remove them periodically with dump-grep-load
>(or remove her account on my machine..).

We're using flex to generate the parsing tables.  AFAIK, it doesn't support 
runtime specified options.  Trimming replicated special characters could be 
implemented pretty easily.  Would you care to write the code?

David