[Rspamd-Users] learn by sending to alias

Kai Schaetzl maillists at conactive.com
Thu Dec 13 14:04:46 UTC 2018

I want to enable users to learn messages. I've setup the recommendation of 
using a sieve operation when moving to/from spam folder. But this works 
only when using IMAP or webmail.

I also want to provide users a way to simply send the spam/ham in from 
their mail program. Like described here: 

What exactly gets learned then? Does it learn headers as well? This would 
learn the wrong headers then. Wouldn't it?
There are several ways how messages get forwarded and how different mail 
programs offer to forward. Users most often do not have a choice of a 
specific format.
Should I put a filter script between the alias and rspamc to make it more 
"digestable" for rspamc? If so, did somebody already create a script for 

I made a simple test by sending a spam message, only the few text lines it 
had, to the alias and it got learned. Then I learned those few text lines 
via web interface and it got learned again. I tried a second time via web 
interface and then it refused as "already learned". So, apparently the 
learning via alias learned different content then what I inserted in the 
web interface. Did it learn header+text in contrast to just text that I 
had copied in?

What's the best practice to get this working most hassle-free?



More information about the Users mailing list