Rick Troxel wrote:
On Wed, 11 Dec 1996, Philip Guenther wrote:
guenther> I have found that when sorting out mail from mailing lists
guenther> into separate folders/mailboxes/whatever, the *best* (opinion
guenther> alert!) thing to 'sort' on is the Return-Path: header. Any
guenther> mailing list set up this decade will send out messages with an
guenther> envelope sender of something like:
guenther>
guenther> procmail-request(_at_)Informatik(_dot_)RWTH-Aachen(_dot_)DE
guenther>
guenther> Most (?) MTAs, and sendmail in particular, will put the
guenther> envelope sender in the Return-Path: header, where it can be
guenther> matched on.
I find that very few of the lists to which I subscribe have this header.
I've had to develop my personal criteria on a per list basis.
Even *THIS* (procmail) list arrives here without a "Return-Path:"
header (you meant "header field" I know). Was it set up in this
decade? BTW, they use sendmail here I'm 99% sure.
This list *does* arrive with:
"Resent-Sender:
procmail-request(_at_)Informatik(_dot_)RWTH-Aachen(_dot_)DE"
and
"Resent-From: procmail(_at_)Informatik(_dot_)RWTH-Aachen(_dot_)DE"
both of which (I think) are applied by the list.
Looking at RFC-822, it seems that "Return-Path:" is added by the
*final* transport system, and may contain routing information
(which implies to me that it might vary from time to time as software
is changed by an ISP). I think it's also optional.
LISTSERV lists seem to insert "X-Sender:" as the original sender, and
"Sender:" as the list name, inserting no "Resent-" headers at all.
I can't tell if this is RFC-conformant, but I'd guess it is, just
since they've been around for 10 years now. So, looking at
"Resent-..." won't always work either.
I've found that if you have access to "From ", it's the most consistent
for *ALL* types of lists. If you can count on "Return-Path:", that
may work for you -- but it's probably bad advice to give, since it seems
to be common not to supply it (I have three accounts, none of which
supply that header field except when it's a *local* return path).
I'll argue against using "Return-Path:" unless you want to depend on
information supplied (or not) at the last stage of transport, rather
than by the mailing list itself. Thus, I agree with Rick and
disagree with Philip (from whom I've learned lots though).
All IMHO, of course.
Cheers,
Stan Ryckman (stanr(_at_)tiac(_dot_)net)