Open Discussion → Detect renaming of folder/buckets
In the context of all subfolders of INBOX corresponding to buckets
Renaming folders might be handled like this:
Quick and dirty, just stop using old bucket, and start a new bucket, retraining on everything currently in the bucket.
For extra credit, POPFile might be able to detect that many of the messages in the renamed folder are the same as those in the old folder, so POPFile might be able to create a hypothesis that allows the old training to continue being used for the new renamed bucket/folder. I would probably try to create a featured set whose indexes correspond to some checksum of features that IMAP doesn't mess with - From, To, Subject, contents - and then comparing, looking for common bits in the smaller. But POPFile probably doesn't maintain that much per-bucket state.
Or perhaps run messages in the new folder against all of the old classifier state for the old buckets, and if more than X% match, decide that you have detected a rename.
This is rather how GIT tries to detect renaming of files - i.e. it doesn't, it detects content showing up elsewhere.