[spambayes-dev] An alternative to spambayes.el for those using Gnus

Ted Zlatanov tzz at lifelogs.com
Tue Nov 14 16:14:32 CET 2006


On 11 Nov 2006, f.rougon at free.fr wrote:

> I've been running my own interface code between Gnus and Spambayes for a
> while, and improved it a bit today to the point that I think it should
> be ready for public consumption.
>
> It can do the same things as spambayes.el, but in a way that should be
> cleaner and slightly faster (using `call-process-region' instead of
> `shell-command-on-region', for instance). It also provides a few more
> things, most notably:
>
> - a command for (re-)running the classifier on an article (or
> process-marked articles). Useful when you've recently trained
> Spambayes and want to see how the newly-trained filter
> performs---and maybe even respool some articles with this new
> filter.
>
> - a command to examine what the Spambayes filter thinks of an article
> (read-only operation): whether it is classified as ham or spam, the
> overall spam score as well as the various spam clues with their
> respective scores (from the 'X-Spambayes-Evidence' header).
>
> This has been tested with GNU Emacs 21.4, Spambayes 1.0.3 and No Gnus
> v0.6 (also with Gnus v5.10.7).
>
> It works well for me, and I hope others will find it useful.

Hi,

would you consider merging your code with the Gnus spam.el system?

You need to write a backend, which includes:

- a spam/ham check function (1 function)
- spam/ham register/unregister functions (4 functions)

Plus update several variables.  It's not a lot of work.  Let me know
if you are interested.  Your spambayes.el can exist with spam.el
(exporting functions for its use) or you can merge the code right in.
It's up to you.

spam.el does much of the infrastructure you mention above, especially
deciding when to run the classifier and on which articles.

Thanks
Ted


More information about the spambayes-dev mailing list