Red Hat Bugzilla – Bug 63842
Subject line mangling is turned on by default.
Last modified: 2008-05-01 11:38:02 EDT
The default for new lists is to turn on the misguided addition of [LISTNAME] to the subject of every mail sent out, ostensibly for the purpose of allowing people to filter list mails. Filtering on anything but the SMTP reverse-path is unreliable and will give false positives. Adding crap to the _Subject_ line simply serves to hide the real information that the original sender put there. This option should be off by default. The behaviour of reply_goes_to_list option is far saner - it is turned off by default and the help text explains why it's a stupid idea to turn it on. The subject_prefix option should be handled similarly - not only should it be empty by default for new lists, but the help text should explain why it is undesirable. Something along the lines of... "This text will be prepended to subject lines of messages posted to the list. By adding noise to the subject lines, the real information therein is made more difficult to read, and therefore it is <I>strongly</I> recommended that this is left blank for most mailing lists. <P> Some people have been known to use the addition of the list name to the subject line as a way of filtering list mails. This practice is unreliable, and far better filtering can be achieved by checking the SMTP reverse-path (usually the Return-Path or Sender headers in the delivered mail)."
Defaults are in the realm of religion, some people like the default the way it is. My general inclination as a package maintainer is not to change the upstream sources unless there is a compelling reason and the upstream sources have this on by default. Given that it is trival to change the default in mm_cfg.py to your own individual preference its hard for me to see the choice of a default as a bug. As it turns out I'm in your camp, I happen not to care for this default either, its just that this isn't a bug IMHO. BTW, mailman now adds List-Id to the headers which is an excellent way for users to filter.
I'll accept the argument that we shouldn't change the default, albeit grumpily -- leaving it on by default means that admins who _don't_ think about it will leave it on, rather than only those who are dim enough to actually think it's a good idea. But I certainly wouldn't advocate filtering on List-ID. That has false positives when a mail _originally_ sent via a list is received via other means -- either by _another_ list which doesn't add its own List-ID or more likely if an important mail is bounced _directly_ to a recipient known not to be paying much attention to the list at the moment. If I know you're on vacation and you're going to read the list with the 'd' key on your return, and I see a mail on the list and hit 'b' in pine to bounce it to you so it ends up in your inbox and you actually _see_ it... Does it end up in your inbox and actually get read, or does it hit a false positive because you're filtering on List-ID: and still end up in the list folder with the original? The only reliable filtering method that I've found is the SMTP reverse-path, usually in a 'Return-Path:' header of the delivered mail.