Bug 2400 - fix bayes thresholds in Conf.pm
Summary: fix bayes thresholds in Conf.pm
Status: RESOLVED FIXED
Alias: None
Product: Spamassassin
Classification: Unclassified
Component: Learner (show other bugs)
Version: SVN Trunk (Latest Devel Version)
Hardware: All All
: P1 normal
Target Milestone: 2.60
Assignee: SpamAssassin Developer Mailing List
URL:
Whiteboard:
Keywords:
: 2391 (view as bug list)
Depends on:
Blocks:
 
Reported: 2003-09-02 06:03 UTC by Al Smith
Modified: 2003-09-03 02:34 UTC (History)
1 user (show)


Attachment Type Modified Status Actions Submitter/CLA Status
fix patch None Daniel Quinlan [HasCLA]
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Al Smith 2003-09-02 06:03:04 UTC 
When spamd or spamassassin start up, the value of
bayes_auto_learn_threshold_nonspam is set in Conf.pm to -2.0.

During the $self->init call in compile_now, this value gets
changed to the value contained in share/spamassassin/10_misc.cf


[SourceForge's CVS seems not to be working so I can't double check
that this issue is still in current-CVS, but this discrepancy is
contained in the bleeding edge files available on the SA download page].
Comment 1 Malte S. Stretz 2003-09-02 07:11:14 UTC 
Took me a while to get the point ;-) I guess the value in 10_misc.cf is bogus; 
I dunno that logic but a positive value sounds weird to me. 
 
Hmmm... was that value introcued after 2.5x? If yes, we should maybe rename it 
to bayes_auto_learn_threshold_ham to be consistent (or rename it officially 
but internally keep the old one, too)...
Comment 2 Malte S. Stretz 2003-09-02 07:27:39 UTC 
forgot to set the milestone
Comment 3 Justin Mason 2003-09-02 11:26:07 UTC 
Subject: Re: [SAdev]  wrong value for bayes_auto_learn_threshold_nonspam in 10_misc.cf

> Took me a while to get the point ;-) I guess the value in 10_misc.cf is bogus; 
> I dunno that logic but a positive value sounds weird to me. 
>  
> Hmmm... was that value introcued after 2.5x? If yes, we should maybe rename it 
> to bayes_auto_learn_threshold_ham to be consistent (or rename it officially 
> but internally keep the old one, too)... 

No, the value in 10_misc is right for 2.60.   The value in Conf.pm is outdated,
but is immediately overridden by 10_misc anyway ;)

--j.
Comment 4 Daniel Quinlan 2003-09-02 17:26:07 UTC 
Created attachment 1307 [details]
fix
Comment 5 Daniel Quinlan 2003-09-02 17:27:17 UTC 
*** Bug 2391 has been marked as a duplicate of this bug. ***
Comment 6 Daniel Quinlan 2003-09-02 17:27:48 UTC 
change summary
Comment 7 Simon Byrnand 2003-09-02 18:01:28 UTC 
Just checking :) My bug 2391 that was marked as a duplicate was mainly refering 
to a documentation error rather than a code error, does this fix address the 
generation of the man page as well ? (I'm not 100% sure as Conf.pm looks like a 
combination of documentation and code to me :)
Comment 8 Theo Van Dinter 2003-09-02 19:22:05 UTC 
+1 this was my bad, sorry about that. :(
Comment 9 Justin Mason 2003-09-03 09:51:54 UTC 
+1.  we should at some point add some doco to the build procedure so we don't
miss that kind of thing in future (ie. "doublecheck defaults in 10_misc.cf match
up with Conf.pm")
Comment 10 Theo Van Dinter 2003-09-03 10:34:21 UTC 
committed.