Re: bogo support: suggestion, question and patch

>>>Alexander Zangerl said:

 > It Would Be Nice(TM) if exmh would feed spam/ham to the bayesian
 > updater programs in one go, not via stdin and one per exec. i've added
 > that choice.

Thanks!  I've picked up your patch.

 > some people like me reinforce spamassassin's classification by 
 > bulk-updating spam messages that weren't autolearned (which spamassasin
 > does only for highly scoring mails), so i run mark-as-spam on lots 
 > of selected messages at a time. SA is a hog and the less often
 > sa-learn is forked, the better. 

There is also the Bogo_FilterFolder variation on this
that I use.  I basically back up spamassissin with my own procmail
rules, or manually refile things into a junk folder. I then aim
sa-learn at that folder directory.  Is that worth exposing
Bogo_FilterFolder in the menus, too?  Hmm - looks like it
is specific to sa-learn, which is probably why I didn't do that already.
 
 > a patch is attached which adds a pref setting that lets the
 > user choose between stdin feeding and handing pathes to the program.
 > would somebody with CVS access please add that? the doc text
 > needs some extra info about bogofilter and spamoracle: i don't 
 > use these and do not know if they can cope with pathes instead of stdin.
 > 
 > so much for suggestion and patch; here comes the question: after
 > enabling
 > bogofilter processing, exmh must be restarted for the options to work
 > out.
 > if you don't do this, exmh will try to use bogofilter as update program
 > and fail at this. (i tried it a few times just to rule out PEBKAC...)
 > why doesn't this setting become active straight after saving the
 > preferences? 
 > AFAICT most other preferences become active instantly...

I didn't quite follow your description of the error - there may be
some clues in the log about what it is trying to do.
--
Brent Welch
Software Architect, Panasas Inc
Accelerating Time to Results(tm) with Clustered Storage

www.panasas.com
[email protected]