Bug 2469 - SA 2.60 chokes on messages without Message-ID
Summary: SA 2.60 chokes on messages without Message-ID
Status: RESOLVED FIXED
Alias: None
Product: Spamassassin
Classification: Unclassified
Component: Learner (show other bugs)
Version: SVN Trunk (Latest Devel Version)
Hardware: Other NetBSD
: P3 normal
Target Milestone: 2.60
Assignee: SpamAssassin Developer Mailing List
URL:
Whiteboard:
Keywords:
: 2152 (view as bug list)
Depends on:
Blocks:
 
Reported: 2003-09-16 16:32 UTC by Klaus Heinz
Modified: 2003-09-18 13:07 UTC (History)
1 user (show)


Attachment Type Modified Status Actions Submitter/CLA Status
message which triggers the error message/rfc822 None Klaus Heinz [HasCLA]
suggested patch for the "one true way"(tm) patch None Theo Van Dinter [HasCLA]
suggested patch patch None Theo Van Dinter [HasCLA]
fixed a few buglets patch None Theo Van Dinter [HasCLA]
new suggested patch patch None Theo Van Dinter [HasCLA]
Show Obsolete (3) 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 Klaus Heinz 2003-09-16 16:32:02 UTC 
SA 2.60-rc4 and -rc5 both show the following error (I didn't test older versions):

  ...
  debug: bayes: found bayes db version 2
  debug: Score set 3 chosen.
  debug: Initialising learner
  Use of uninitialized value in substitution (s///) at /usr/pkg/lib/perl5/site_perl/5.6.1/Mail/SpamAssassin.pm   line 442.
  debug: Learning Spam
  debug: uri tests: Done uriRE
  ...

Both messages where I saw this didn't have a "Message-ID" which explains, why line 442
  $TIMELOG->{mesgid} =~ s#/#-#g;
doesn't work.

ciao
     Klaus
Comment 1 Klaus Heinz 2003-09-16 16:35:52 UTC 
Created attachment 1388 [details]
message which triggers the error
Comment 2 Theo Van Dinter 2003-09-16 19:35:37 UTC 
this seems very related to my bug 2152 ...

I'm seeing if I can't get a generic patch done up now which doesn't really solve 
2152, but should make M::SA::Bayes->get_msgid() the "one true way"(tm) of 
getting Message-Id WRT bayes...
Comment 3 Theo Van Dinter 2003-09-16 19:43:02 UTC 
Created attachment 1389 [details]
suggested patch for the "one true way"(tm)
Comment 4 Theo Van Dinter 2003-09-16 21:35:48 UTC 
moving to 2.60
Comment 5 Theo Van Dinter 2003-09-16 21:37:38 UTC 
*** Bug 2152 has been marked as a duplicate of this bug. ***
Comment 6 Theo Van Dinter 2003-09-17 13:25:15 UTC 
Created attachment 1393 [details]
suggested patch

after discussing things with quinlan, I hacked through and removed the
message-id passing and such...

The attached message doesn't cause any uninitialized value warnings now, and
properly gets learned (it did before too) with a generated msgid:

debug: bayes: Learned '72c4e142e4f87d1289b6cc9f5215f0af4e9d5d1f@sa_generated'
Comment 7 Theo Van Dinter 2003-09-17 14:09:44 UTC 
Created attachment 1394 [details]
fixed a few buglets
Comment 8 Theo Van Dinter 2003-09-17 20:16:44 UTC 
Created attachment 1396 [details]
new suggested patch

after a discussion with quinlan, here's the new version.  it keeps the old API,
and now allows people to pass a message identifier in which will be used for
learning.
Comment 9 Daniel Quinlan 2003-09-18 18:48:05 UTC 
+1 works in my testing

Also, the documentation at the top of PerMsgLearner.pm makes no sense to me.
Comment 10 Daniel Quinlan 2003-09-18 21:07:05 UTC 
applied, closing