Hey all,

I plan to release a new version of my SpamAssassin Trainer script in early April 2008 that will be in either Perl/CGI (web-friendly) format, PHP format, or a Perl script suitable for setting up as a cron job.

I wanted to get some feedback on an additional idea I had while discussing mailbox cleaning with another user.

A problem arises when you have a domain name with multiple users who don't manage their ham/spam folders very well. By not deleting old messages that have already been scanned, scanning takes longer and longer, and you pollute the bayesian database with old outdated messages (essentially training it that messages arriving with very old dates are acceptable or not). There are obvious privacy/security/legal implications with having the scanning script empty mailboxes in an automated method (losing Emails that need to be kept, for example) once scanning is completed, which is why I removed the ability to do that from my training script a few versions back.

So this new idea is to build in some sort of Email-based notification process that would cause my SpamAssassin trainer script to send an Email once scanning is completed, and have that message tell users to (a) empty their spam folders, (b) save any copies of ham messages they need to retain from their ham folders, and (c) explain why emptying the folders will be a very beneficial help (that NOT emptying the folders would slow the scanning process to the point of being unusable, etc.)

This notification process would obviously be very handy for the cron-friendly version of the script I want to write, and I could even make the message text configurable to the users of the script.

I'm just not sure which of the poll options to build.

I could build in an option to send a message to a single address, suitable for single domain owners who set up multiple addresses for themselves, or to a mailing list address (you'd have to set up a Mailman list on your own)

Alternately, I could send a message to every mailbox detected during the scanning process (handy for domains with many users, but if you have more than 400 users, you could run into issues of being flagged a spammer for notifying all of them in rapid succession).

Or I could build in BOTH options and let you (the admin user installing the script) decide which one to use. Or better yet, maybe let you set both options at the same time to send a message to an admin mailbox plus each individual user detected during the scanning process.

Of course I'd configure the script such that you could turn off both/either option(s) and have no Email sent once scanning is done... if you're a single user manually running the script you'll already know when scanning is completed.

Or maybe neither of those options is a good one (any other ideas?).

I'm open to any suggestions. I'll leave the voting open for the month of March, and build in whichever idea(s) is(are) most voted on.

Thanks!

The SpamAssassin trainer is no longer working well. Lunar Pages have made too made changes for "security."  When is your new script ready?