Floating point errors?

[Ingomar Wesp]

Ingomar Wesp wrote:

> For some reason, when manually marking spam or ham, bogofilter was always
> called with the -N and -S options respectively, even if the message was not
> previously registered at all.

I just decided to investigate on where this misconfiguration came from. The 
results are bad news:

I'm using KMail 1.9.7, which is KDE's default mail user agent. KMail has a 
built-in feature called "Anti-Spam Wizard" that automatically creates filters 
for using external anti-spam software like bogofilter or spamassasin. 
Unfortunately the filter setup that is created for bogofilter looks like this 
(stuff that is irrelevant for bogofilter has been removed):

+----------------------+----------------------------------------+-------+
| Filter name          | Action                                 | Auto? |
+----------------------+----------------------------------------+-------+
| Bogofilter Check     | Pipe through    "bogofilter -p -e -u"  | Yes   |
| Classify as SPAM     | Execute command "bogofilter -N -s"     | No    |
| Classify as NOT SPAM | Execute command "bogofilter -S -n"     | No    |
+----------------------+----------------------------------------+-------+

Obviously, each time the user applies "Classify as SPAM" on a message that has 
not previously been registered (either because it's an old message that has 
not been piped through bogofilter before or because bogofilter was unsure 
about whether the message was ham or spam), the ham values for all tokens in 
this message (and the .MSG_COUNT) get decremented. Which is bad, because they 
were not wrongly incremented in the first place. The same applies 
for "Classify as NOT SPAM" and spam-counts respectively.

If I'm not mistaken, this is a bug in KMail and should be corrected. In case 
there's no one else with a better grip on the English language who wants to 
do it, I'll be filing a bug report soon.

Best regards,
Ingomar Wesp