Dear eric, We would like to use the toaster in an ISP environment.... So our User shuld be able to set their whitelist, black list, and spam filter sensibility...
Like with http://sourceforge.net/projects/webuserprefs/ So they can manage their incoming messages or per domain or per user... But I am a little confused which is the best solution.. Probably is Per user with your suggestion below.. with then spam button in squirrelmail that train bayes.... But... :( In my opinion maybe is better to put all SpamAssassin user prefs in mysql... But in this case I do not know how to retrive them with the toaster.. Should I use spamc instead of spamd in /var/qmail/supervise/spamd/run ??? This is mine... exec /usr/bin/spamd -x -u vpopmail -s mail 2>&1 I need to change it with appropriate settings to spamc?? :-(( Thanks.. (very confused please help ) -----Messaggio originale----- Da: Eric "Shubes" [mailto:[EMAIL PROTECTED] Inviato: martedì 27 febbraio 2007 16.39 A: qmailtoaster-list@qmailtoaster.com Oggetto: - Re: [qmailtoaster] spamassassin per user Alberto Catagni wrote: > Hi eric, > > first of all thanks for the great work that you and your tema are doing > on the qmail toaster... > > I am writing to ask if there is some white paper on how to implement > spamassassin with a per user preferences... Not that I'm aware of. The best toaster related SA documentation that I know of is at http://wiki.qmailtoaster.com/index.php/Spamassassin (and it could use some work). That being said, what sort of user preferences are you looking for? SA can be enabled/disabled per user by adding an entry in the simcontrol file for the specific user(s). There is also the 'spambox' option for qmailadmin which is selectable by user. I know that some users have put together varying methods of using sa-learn to train the bayes filter, but I don't think that is exactly "per user". I think a few have also used MySQL for the bayes database, but I don't believe that was 'per user' either. I think that a per user implementation of bayes using MySQL would be a nice feature, but it's not in the works that I know of. IMHO, bayes is best suited to filtering at the user level, where the end user can make the call on what is spam and what is not. That's generally done by the client instead of the server. It doesn't mean that the server *couldn't* provide user level bayes filtering though. However, it would be necessarily limited to an IMAP configuration. > I saw this link http://www.qmailtoaster.com/old/spam/ This link is ancient history (according to the dates). > But I guess if you are thinking on a new upgrade of the toaster with > this features... > > > > Waiting 4 your answer... > > Thank you... > > Alberto > -- -Eric 'shubes' --------------------------------------------------------------------- QmailToaster hosted by: VR Hosted <http://www.vr.org> --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- QmailToaster hosted by: VR Hosted <http://www.vr.org> --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
