Hello,
I have subscribed to this mailing list primary to share my wishes
for improvements in spamassassin, and get feedback on them, before
submitting them in bugzilla.
I run an MTA with 15 000 addresses. 2000 of them are local, and the
rest are aliases. Some users have more than one redirect, e.g. munich@
and muenchen@ refer to the same user. I want to offer good service and
that's why I gave the users the possibility to edit their spamassassin
preferences.
Except for local addresses, most users don't use the results of
spamassassin. Hence I would appreciate if there is an option for the
user, that disables the filtering, or even better, global setting
(on/off), which can be ajusted by each user personally. In this way I
can finally escape from the spam attacks causing DOS.
From time to time users are deleted. In such cases I would like to
have an easy way to remove the user's settings. Deleting the databases
manually, or reading local.cf, is not really very nice. It would be
better if a command like sa-user --remove 'user1' deletes the settings
for a user, and calls sa-learn --clear 'user1'.
Often a mail is comming whose envelope contains more than one
recipient. In such cases the preferences for the default-user shall be
taken into account. However, as I said, the default-user must prove in
some cases if the mail is spam, but it could happen, that if a mail is
going only to user1 and user2 and both have disabled their
spam-filtering, then the mail does not need to be filtered. This
applies also for other settings. Therefore I would suggest if the
settings of all involved users in a mail are read and "the least
restrictive" one is choosen. E.g. if user1 and user2 have the same
adjustment, it is always taken into account. If user1 has
required_score 5 and user2 has required_score 6 and the site-default
is required_score 7, then a mail with 6 scores shall be announced as
spam for both users. And so on for each setting.
From time to time, users want to rename their address. In such
cases I would like to have an interface like sa-user --rename user1
user2, which transferes all settings (and bayer knowledge) from user1
to (the new) user2.
And as I said some users have more addresses, but they want to keep
the same settings and bayer-knowledge for all of them. For such cases
it would be like to have user preference like
equivallent_user user1 user2
meaning that if a mail is coming to user1 then the settings of user2
shall be taken into account (and this implies also the bayes knowledge).
Thanks for your attention,
Дилян