You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@spamassassin.apache.org by Justin Mason <jm...@jmason.org> on 2004/11/24 00:00:57 UTC

Re: proposal: 3.1 release target

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


sounds good to me.  One thing - could you run some tests on the sampling
so we can see how reliable it is, in terms of hit-frequencies?   I'd
like to get a "sanity check" on that, it's a key aspect.

- --j.

Daniel Quinlan writes:
> I think we should define our release target date and try to be
> aggressive due to the memory issues in 3.0, the easier process with the
> new mass-check, and the overall accuracy and speed improvements we've
> already made are worth getting out.
> 
> To recap, 3.0.0 was released on 2004-09-22 and 3.0.1 was a month later.
> If we target for last major release + 4 months (very aggressive and
> unrealistic), that puts us at 2004-01-22.  3.1 is going to have
> single-cycle mass-check via the sampling (now checked into SVN), so we
> should be able to do that bit in 2 weeks at most.  Also, if we only put
> 4 months of code improvements into the tree, our pre+rc phase might only
> take a month.
> 
> So, my proposal is simply that we target January 22nd for the first
> pre-release.  I think we could have a release out by mid-February.
> 
> We'll probably need to leave some speed and efficacy improvements out of
> 3.1 unless some major patches are contributed, but I think we could just
> keep the same goals in place for 3.2 and do another fast cycle.
> 
> Daniel
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Exmh CVS

iD8DBQFBo8EpMJF5cimLx9ARAqI5AJ4mla9YCKn1uy60GyC6/bex0Gu9iwCgiZeL
QnloLPaBjpzqxGrwUSoQyak=
=fl0A
-----END PGP SIGNATURE-----


Re: proposal: 3.1 release target

Posted by Daniel Quinlan <qu...@pathname.com>.
Justin Mason <jm...@jmason.org> writes:

> sounds good to me.  One thing - could you run some tests on the
> sampling so we can see how reliable it is, in terms of
> hit-frequencies?  I'd like to get a "sanity check" on that, it's a key
> aspect.

Yes.  I will do so.  I'm going to compare auto-learning with
sample-learning using a similar percentage of messages learned.  It
should be easy enough to get it to a similar error rate if it's too
good.

Another thing I want to do is make it deterministic instead of using
rand(100).  If I change it to "learn 1 in N" instead of a percentage,
then I can easily do a mod on the md5sum of the id and/or date.

Daniel

-- 
Daniel Quinlan
http://www.pathname.com/~quinlan/