I'm running Gnus v5.10.7 and using spam-stat.el via spam.el for spam
filtering.

My setup is as follows:

Mail is split according to senders, then if none matches the spam
splitter is used. nnml:raw.spammy is my spam split destination; all
the rest goes to nnml:raw. These 2 groups are unclassified (neither
spam nor ham).

I have a single group classified as spam: nnml:spam.explicit. This is
my spam process destination.

All other groups are classified as ham.

Currently my workflow is as follows: enter the /nnml:raw\(\..*\)?/
groups, move all the ham articles into nnml:misc to be read or manually
split later, then mark all the rest of the articles as spam and exit the
group, causing them to be processed and moved to nnml:spam.explicit.

Now (correct me if I'm wrong) I could have /nnml:raw\(\..*\)?/
classified as spam groups for the same basic effect. However I don't
like that I lose all the score highlighting when they are auto-marked as
spam, and I'm nervous about the default mark being "spam". Please tell
me if this is a legitimate concern.

I'm thinking what I will do is implement an interactive function which
overrides the ! key in the raw groups. It will remove all marks and
move the article into nnml:misc.

Is there a reason that gnus-ham-process-destinations only works for spam
groups and not in non-ham groups? If the registry is used that should
avoid double-processing concerns, no?

--
vsync
http://quadium.net/~vsync/

The beauty of Amendment II is that it protects the ability to deal
with both a tyrannical government and a zombie apocalypse.