You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@spamassassin.apache.org by bu...@bugzilla.spamassassin.org on 2004/01/06 23:44:07 UTC

[Bug 2751] Argument isn't numeric

http://bugzilla.spamassassin.org/show_bug.cgi?id=2751





------- Additional Comments From gordon@cs.ualberta.ca  2004-01-06 13:49 -------
I believe we are experiencing the same bug.  It did not show up until I tried to
change the status of some messages from spam to ham.  Now I can get it to happen
just by toggling a single test message back and forth from spam/ham.

We were running SA v2.55
We upgraded to 2.61, installed DB_File-1.807, perl, v5.6.1 built for
sun4-solaris-64int

Each user has their own .spamassassin directory (UNIX environment).
I've ran 'sa-learn --import' and then proceeded to give sa-learn my most recent
missed spam for it to learn, followed by some spam and ham folders (which
theoretically should be redundant).  Then I noticed that I had some
false-positives in the spam folders so I weeded them out into a separate ham
folder and fed that into sa-learn.  At that point I started getting these errors.

So I have now moved my entire .spamassassin folder aside and started from
scratch.  This is not a solution I want to proffer to my 300+ users.

The offending piece of code is in
/usr/perl5/site_perl/5.6.1/Mail/SpamAssassin/BayesStore.pm
Subroutine tok_put, last if statement:
    my $oldmagic = $self->{db_toks}->{$OLDEST_TOKEN_AGE_MAGIC_TOKEN};
    printf STDERR "atime = $atime, oldmagic = $oldmagic\n";
    if (!defined ($oldmagic) || $atime < $oldmagic) {
      $self->{db_toks}->{$OLDEST_TOKEN_AGE_MAGIC_TOKEN} = $atime;
    }

NB:  the printf is mine, to see what might be the problem.

Sample output:

sa-learn -D --ham --mbox mspam
...
debug: Removing Markup
debug: Learning Ham
debug: uri tests: Done uriRE
debug: 200401061749.i06HnWto027946@methan.in-berlin.de: already learnt as opposi
te, forgetting first
debug: uri tests: Done uriRE
debug: tokenize: header tokens for *m = " 200401061749 i06HnWto027946 methan in
berlin de "
debug: tokenize: header tokens for *F = "U*Sendingout D*zccn.net D*net"
debug: tokenize: header tokens for To = "U*plan D*gate.bitrot.de D*bitrot.de D*d
e"
debug: tokenize: header tokens for Mime-Version = "1.0"
debug: tokenize: header tokens for *c = "/plain; charset="ISO-8859-1""
debug: tokenize: header tokens for *R = "U*Sales D*help-boost.com D*com"
debug: tokenize: header tokens for Content-Transfer-Encoding = "8bit"
debug: tokenize: header tokens for *p = "U*plan-owner D*bitrot.de D*de"
debug: tokenize: header tokens for X-Orcpt = "rfc822;gordon@cs.ualberta.ca"
debug: tokenize: header tokens for *r = "  zccn.net ([218.86.53]) by methan.in-b
erlin.de (8.12.10/8.12.10/Debian-4)         <pl...@bitrot.in-berlin.de>; "
debug: tokenize: header tokens for *r = "  zccn.net ([218.86.53]) by methan.in-b
erlin.de (8.12.10/8.12.10/Debian-4)         <pl...@bitrot.in-berlin.de>;    metha
n.in-berlin.de (methan.in-berlin.de [160.45.10]) by hirsch.in-berlin.de (8.12.10
/8.12.10/Debian-6)         <pl...@bitrot.in-berlin.de>; "
atime = 1073418626, oldmagic = 20700626^?��?
Argument "" isn't numeric in numeric lt (<) at /usr/perl5/site_perl/5.6.1/Mail/S
pamAssassin/BayesStore.pm line 1249.
atime = 1073418626, oldmagic = 20700626^?��?
Argument "" isn't numeric in numeric lt (<) at /usr/perl5/site_perl/5.6.1/Mail/S
pamAssassin/BayesStore.pm line 1249.

I tried running --rebuild, I tried --import 'ing again.  It would be nice if the
rebuild function would at least fix the problem, even if its not known what
causes it.

G.H.A.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.