Some code for interacting with spamprobe from within Rmail. This code assumes that you have a functional mailsystem with procmail calling spamprobe and inserting an X-SpamProbe header in filtered messages before the messages reaches RMAIL.
I wonder: When you run spamprobe train within procmail, and it misclassifies an email, and you then use your dwim defun to reverse the decision, doesn’t that just cancel what it learnt previously? Perhaps you have to delete the old info and add the new, instead of just adding the new. – AlexSchroeder
No, telling spamprobe that tokens previously learned as spam is in fact good or vice versa does more than just just removing them from the wrong database, it will also add them to the correct one. The reason I decided to run the message through procmail again is in fact laziness. It was easier to do than to decide which babyl file it really oughta be in and move it there by way of lisp code. – HenrikEnberg
I tried to use your package and it does not work here. When trying to debug, I found that it never catches the spamprobe header. – XavierMaillard