Differences between 0.13.0 and 0.13.3 [was: upgrading ...]

[David Relson]

At 10:19 AM 5/29/03, Shawn Barnhart wrote:

> > > Also, there have been a number of minor defects fixed, so I'd recommend
> > > getting 0.13.3 first.  Then, retrain.
> >
> > Will 0.13.0 kill me?  I'm building it out of FreeBSD ports and ports only
> > has 0.13.0 currently.
> >
> > Can you describe the bugs fixed with 0.13.3?  I'll submit a PR and maybe
>it
> > will get updated.
>
>Hmm, looks like there's an open PR for updating to 0.13.2.1 from May 24 --
>no commits due to a ports freeze waiting for 5.1-RELEASE.

Shawn,

Most of the changes don't affect the program itself.  The ones that do are 
in charset.c, lexer.c, and main.c.  If you don't use 
replace_nonascii_characters or bulk mode, the changes to charset.c and 
main.c won't affect you.  The "scanner reject" fix in lexer.c affects very 
few messages (2 or 3 of the 50,000 in my corpus).  It'll be obvious when 
you encounter the problem.

Below is a brief summary of the changes.

David

0.13.0 to 0.13.3 - change summary

bogofilter.cf.example - removed some extra hash marks
bogofilter.spec.in - changes for staticly linked binaries
                      new file doc/integrating-with-postfix includes
bogolexer.c - added "-V" (version) option
charset.c - "replace_nonascii_characters=yes" no longer overrides
             "ignore_case=no" option
lexer.c - better checking for extra long tokens so that scanner won't
           quit with "input buffer overflow, can't enlarge buffer
           because scanner uses REJECT" message
main.c - output bulkmode file names to stdout (or -O, if used)
Makefile.am - changes for statically linked binaries
rstats.c - if "stats_in_header=no", include blank line before histogram
token.c - code changes to quiet gcc-3.3 warnings