Description of problem: Noticing that spamassassin didn't seem to be checking some incoming mail at all, I moved that mail to a separate folder and ran sa-learn on it. Sa-learn appears to be crashing when processing some of these messages. Version-Release number of selected component (if applicable): 2.63-0.2 How reproducible: Every time Steps to Reproduce: 1. run sa-learn on a folder of spam 2. 3. Actual results: sa-learn crashes with a segmentation fault Expected results: sa-learn processes the messages Additional info:
Created attachment 98509 [details] Debug output of crashing sa-learn (with mail address censored)
It would really help for the developers to debug and fix this if you can attach an isolated mbox file that causes the crash. See if you can cut pieces out that are unrelated to the crash, as well as anything for privacy reasons.
I did talk with a couple of developers on IRC. In the end, I deleted my SA installation and started from scratch. That seems to have solved the problem. It's likely that the Bayesian database had become corrupted. Mail was becoming unusable again so I couldn't wait any longer for a fix. If this reappears, I'll reopen the bug.