Pan 0.145 from git, compiled on Debian stretch. Same applies to the Debian version.
I'm downloading headers from a group where one poster has posted millions of multi-part articles, but only one part of each.
I have a rule scoring these articles -9999 and Delete articles scoring at: -9999. It works. The posts do not appear in the Group pane, nor are they saved in the .pan2/groups/ file.
Even with the scoring rule, Pan is using a lot of memory on these.
Headers Received: 5254001
Network Received: 2.3GB
Pan Memory Usage: 10.1GB
If I stop Pan here, the groups file is only around 25MB. If I turn off the scoring rules and run it again it's still only about 1.3GB. About 3.5 million of the 5.25 million articles are from the one poster.
This same issue that was reported around January 2017, but without details.
I've downloaded headers for 1 billion parts, more than 20 million multi-part articles, in the same group, prior to this but for some reason these just kill Pan.