Bogofilter, script segfaults

[David Relson]

On Tue, 28 Jun 2005 12:37:19 -0700
Rodney D. Myers wrote:

> I'm running debian Sarge, and bogofilter version
> bogofilter -V
> bogofilter version 0.95.1
>     Database: Sleepycat Software: Berkeley DB 4.3.28: (April 22, 2005)
> AUTO-XA Copyright (C) 2002-2005 Eric S. Raymond,
>     David Relson, Matthias Andree, Greg Louis
> 
> When running my script to train spam/ham, in known folders I get the
> following errors;
> 
> Tinkering with the SPAM
> # 244833 words, 12425 messages
> free(): invalid pointer 0x8107a98!
> /home/rodney/bin/spam-filter: line 9: 18857 Segmentation fault
> bogofilter -svvv -B /home/rodney/Mail/spam/
> 
> 
> In line 9, I have the following;
> 
> bogofilter -svvv -B /home/rodney/Mail/spam/
> 
> The only thing I can see with that. Is that this particlualr folder
> holds 12425 messages.

Hi Rodney,

Interesting !  The only code difference of significance is in the
processing of RFC2047 tokens, which are now converted to unicode (if
that's in effect).  Likely the problem is there!

Is there any chance you can isolate the problem to a single message, or
a mbox file with only a few messages in it, and send that to me?  Not
as good, but still helpful would be to run within gdb and send a
backtrace when the problem occurs.

Regards,

David