SmartList version 3.15 will be released on or about the 30th of August. The delay is in the hopes that someone running SmartList under either postfix or qmail can test something for me and confirm whether the fix in the prerelease is sufficient.

What's the problem that needs testing, you ask? The documented procedure for providing both a digested and an undigested version of a list is to create them as separate lists, subscribe the digested list to the undigested list, set the "undigested_list" variable of the digest to point to the undigested list, and finally make the undigested list be the only address in the (delinked) digest list. As a result, as post from a subscriber to the digested list typically goes first to the digested list, which then forwards it to the undigest list. When it sends it out, it sends it back to the digested list where it will then end up in the next digest. The problem is that on systems that use the Delivered-To: header to prevent mail loops (qmail & postfix), this double-reception of the message by the digested list is detected as a mailloop and is suppressed, resulting in the message never being posted to the digest list.

My fix is for the digested list to remove any Delivered-To: headers that contain its own address before forwarding the message to the undigested list.

My proposed fix can be found in the rc.submit file from the prerelease. Can someone on a system running postfix or qmail please try this out and tell me whether my fix works? I've thrown several messages through the recipes and they appear to work, but I would feel better about such a late-breaking change (I only learned about the bug on the 23rd) if someone in the situation could test it.

Philip Guenther Procmail & SmartList maintainer bug@procmail.org