Precedence Headers
Marc A. Donges
I'm looking for a precise definition of the Priority and Precedence
Headers. It was not in RFC 822 and I couldn't find anything in the
rfc-index or via AltaVista. Does anybody know a source for the
information I need?
Marc
--
_ _ Marc A. Donges +49 791 51804
'v' <ma...@yodanet.schwaebischhall.de>
/ \ PGP-Key(RSA): 1024/46D29AE5
W W Fingerprint: 12C0 85FC 8F4C 7763 417D FB8B 6736 FFAB
era eriksson
<filter.a...@yodanet.schwaebischhall.de> posted to
comp.mail.headers:
> I'm looking for a precise definition of the Priority and Precedence
> Headers. It was not in RFC 822 and I couldn't find anything in the
> rfc-index or via AltaVista. Does anybody know a source for the
> information I need?
Dan Bernstein's pages (<ftp://koobera.math.uic.edu/www/proto/immhf/>
-- definitely bookmark stuff) mention RFC1327 for the Priority: header
but nothing at all for Precedence:. (RFC1327 is not very meaningful,
but it lists some valid values.)
In practice there are three precedence levels in common use;
$ grep '^P[a-z]*=' /etc/mail/sendmail.cf
Plist=-30
Pbulk=-60
Pjunk=-100
In addition, stuff like "first-class" and "normal" is frequently seen,
but they are (commonly) treated as if there was no Precedence: header.
The following is just off the top of my head but hopefully good
enough:
list -- mailing list messages
bulk -- other bulk postings
junk -- other automatically generated messages, such as
autoresponder responses, bounce messages, etc
The numbers in Sendmail's configuration file indicate a processing
priority for these messages, where zero means most preferred. (This
should be reserved for one-to-one human communications.) I believe
(but have never even attempted to verify) that on most modern systems,
this setting doesn't play any significant role under normal
conditions.
Hope this helps,
/* era */
--
.obBotBait: It shouldn't even matter whether <http://www.iki.fi/era/>
I am a resident of the state of Washington. <http://members.xoom.com/procmail/>
* Sign the European spam petition! <http://www.politik-digital.de/spam/en/> *
Alan J Rosenthal
[mostly good stuff, but I disagree with:]
>list -- mailing list messages
>bulk -- other bulk postings
In my experience, "bulk" is more commonly used for mailing lists than "list",
silly as that is. What I'm saying is that if you are using the Precedence
header to detect mailing lists (e.g. not to send vacation notices),
you should definitely include "bulk" amongst the possible-mailing-list
indications.
In a quick scan of actual messages lying around my accounts, I found
"Precedence: list" headers only on the old best-of-security mailings and
the redhat-watch-list, and "Precedence: bulk" headers on almost all other
mailing lists. (I'm sure there are others which use "Precedence: list",
but they seem to be in the minority.)
Strangely, one message said "Precedence: auto_reply", which is at best
useless, if not worse than useless...
>The numbers in Sendmail's configuration file indicate a processing
>priority for these messages, where zero means most preferred. (This
>should be reserved for one-to-one human communications.) I believe
>(but have never even attempted to verify) that on most modern systems,
>this setting doesn't play any significant role under normal conditions.
This doesn't contradict what Era Eriksson said, but note that the values
of the Precedence field are often interpreted by autoresponders such as
vacation programs, and should be. Mailing list software should insert
"Precedence: bulk" and vacation programs should fail to respond to messages
bearing Precedence values of either "bulk", "list", or "junk" ("junk"
being inserted by lots of autoresponders, including vacation programs).
The vacation message itself should insert the header "Precedence: junk"
or perhaps "bulk".
This is probably not the best design, but it's probably the most
interoperable. (Of course there are other things vacation programs should
do to avoid replying to mailing lists, such as insisting upon finding the
addressee in some headers.)