Ken Hooper proposed,
| This is precisely the sort of thing
| SmartList is good at; look in the body and try to decide intelligently
| whether Reply-To should be set off-list ("I don't subscribe to the list")
| or on- ("Please don't CC me"), then set it. This is a fairly trivial regexp
| problem and an extremely trivial SmartList problem.
Not a good idea, Ken. No automated parser is going to be smart enough to
make the right choice every time. For starters, you have to be sure that
it isn't fooled by discussion of the ideas. Your post contained both those
contradictory phrases (let's assume that the parser, in reading *this* post
from me, would ignore them because of the citation characters) and you didn't
necessarily mean either of them; would a parser be able to deal with that?
And what happens, moreover, when people misspell or rephrase those standard
terms? The SmartList aspect may be trivial, but the regexp facet is very
complex. There's a reason that the diversion regexp in SmartList is more
complicated than simply "(un)?subscribe|help|get|list".
Here's a suggestion, Ken: try it in your own .procmailrc for your own sub-
scriptions. See how bullet-proof you can make the regexps, and be sure to
use formail -i"Reply-To: your software's selection" rather than -I so that
you can see whether there was an incoming Reply-To: and compare the result
to it.
List software should not make automated decisions that override the poster's
supplied return address. It's one thing if a respondent, knowing what he or
she is going to say in reply, writes to a different address, or if a modera-
tor decides to change it on an occasional article (and that, of course, only
on a moderated list), but those are considered determinations made by human
minds. This is one of those few tasks best left to sentient beings.