bogofilter -- Fast Bayesian Spam Filter / News: Recent posts

The time has come to prepare another bogofilter release to get the bug fixes and new data base support rolled out. Bogofilter can now also be used with KyotoCabinet (code contributed by Denny Lin), LMDB (code contributed by Steffen Nurpmeso). There have been several major bug fixes, many minor bug and a few efficiency fixes, polishing, a license change for the security announcements (for easier distribution they now permit dual-licensing per the GPL v3 or newer), and a bit of maintenance under the hood. The repository has been converted from SVN to Git.... read more

The Bogofilter repository has been converted to Git (from SVN) today.
This sourceforge.net repo is a mirror of https://gitlab.com/bogofilter/bogofilter.

Bogofilter 1.2.0 has been promoted from "current" to "stable" status.

Bogofilter v1.2.0 adds 3 new options to force bogofilter to use a specified number of tokens when scoring a message. The options are:

--token-count=n
--token-count-min=n
--token-count-max=n

When one or more of these options is specified, bogofilter tries to use the specified number of tokens when computing a message's score. Under certain circumstances when multiple tokens have identical scores, bogofilter will compute a score using more tokens than specified.

Fixes to argument processing for bogofilter, bogoutil, and bogolexer.
Minor updates to bf_compact, self tests, and bogofilter-milter.pl.

bogofilter v1.1.7 has been released. It has parsing fixes for bogoutil and bogotune config files and an updated bogofilter-milter.

Adds support for TokyoCabinet database.

Fixes several minor parsing problems.

bogofilter-1.1.2 has a FAQ update and fixes for 2 parsing problems that can cause seg-faults.

bogofilter 1.1.1 has new features for setting min and max token length and scoring based on multi-word tokens. See the RELEASE NOTES for more info.

In the two weeks since bogofilter 1.0.3 was released as "current" package, no serious problems were found. Bogofilter 1.0.3 has therefore been declared stable, and the files been moved to the "bogofilter-stable" package.

A known issue with 1.0.3: the file "RELEASES" that shipped with this release is in fact outdated and shows 1.0.2 as the recommended version. Please disregard this inconsistency.

Greetings,

I have disabled the bogofilter forums on the sourceforge.net site. They have not been used much, and much more timely answers have been provided by the mailing lists.

See http://bogofilter.sourceforge.net/faq.shtml#lists for details on the available mailing lists.

bogofilter 1.0.3 brings assorted bogotune fixes,
improvements to the static RPM building process (as used by the maintainers) and a GNU make compatibility fix.

If bogofilter 1.0.2 works well for you and/or you aren't using bogotune, you can stick with 1.0.2.

Bogofilter-1.0.2 has been promoted from current to stable status.

fixes has some minor configuration script scripts, some minor option errors, suppresses multiple messages when the database nears its maximum size, has updated emacs VM support and an emacs VM entry in the FAQ

Minor bug fixes, most notably for SQLite support code. Binary rpms renamed to reflect supported database (DB4.2.52 and SQLite3.2.8) and link state (shared library vs. statically linked)

bogofilter v1.0.0 has been released!

Prime candidate for the 1.0.0 release :->

Test now!

Fix configuration problems, specifically a problem with comments in config file lines and a problem when an invalid charset is specified.

Minor cleanups to code and of contrib directory.

Ends parsing of mime binary attachments, i.e. applications, audio, image, and video files. Fixes a rarely occurring problem when converting to unicode.

release 0.96.2 has been promoted from Current to Stable status.

Provides unicode support. See RELEASE.NOTES-0.95.0 for details.

Fixed a database recovery problem when Berkeley DB is used in transaction mode.

This release fixes database recovery when the Berkeley DB environement is in PANIC mode. It's relevant to transaction mode users.