From: Russ Allbery (rra@stanford.edu)
Date: Tue Sep 26 2000 - 22:18:55 CDT
Charles Lindsey <chl@clw.cs.man.ac.uk> writes:
> 6. Optional Headers
> 6.1. Mail-Copies-To
> The Mail-Copies-To header indicates whether or not followups to the
> article should be emailed to the poster in addition to being posted
> to Netnews and, if so, may provide an address to which they should be
> sent. The content syntax makes use of syntax defined in [MESSFOR],
> but subject to the revised definition of local-part given in section
> 5.2.
> Mail-Copies-To-content = copy-addr / "nobody" / "poster"
> copy-addr = mailbox
> [<mailbox> allows only a single address. I might alternatively have said
> <From-content> allowing several. Note that the default is in any case
> Reply-To, which in general allows several. John Moreno provided the
> following rationale for this:
> The reasons for allowing an address are to allow the poster to specify a
> special address as a destination for copies of posted replies and to
> allow people posting with an invalid address to specify a valid address
> to which to send a Cc. Only one address is allowed, so as to limit the
> potential for abuse.
> I am not totally convinced, especially about the "abuse" bit.]
Neither am I, but I also don't think anything of interest is gained by
making it From-content. Reply-To has way more abuse potential than this
does, though (think test autoresponders and the like), and since we have
no intention of fixing that, I don't see this as much of an issue.
> NOTE: Some existing practice uses the keyword "never" in place
> of "nobody" and "always" in place of "poster". These usages are
> deprecated. Posting agents MUST NOT generate them, but followup
> agents SHOULD observe them.
> [I stuck my neck out a bit there, by leaving them entirely out of the
> syntax. Is that agreeable?]
Yes, but I think that should be "followup agents MAY observe them." I
don't think it's a good idea to put SHOULDs in notes, and I don't think
it's a good idea to require at the SHOULD level recognition of something
that was never standardized and never will be standardized. The MUST NOT
in this note isn't actually a separate requirement, just a reiteration of
what the BNF says, since mailbox doesn't allow "never" or "always".
> A followup agent MUST NOT (unless its user explicitly overrides it)
> email a copy of the followup article to the author of the original
> article if the Mail-Copies-To header contains the "nobody" keyword.
> If the user explicitly overrides this provision, the followup agent
> MUST/SHOULD/Ought to issue a warning to the user and ask for
> confirmation.
> [MUST/SHOULD/Ought? This hardly causes interoperability problems.
> Perhaps that sentence is not needed at all. Opinions?]
I vote for just dropping that sentence; I think the important idea there
is implicit in the requirement for an explicit override.
> A followup agent SHOULD (unless its user explicitly overrides it)
> email a copy of the followup article to the author of the original
> article if the Mail-Copies-To header contains a copy-addr or the
> "poster" keyword. The copy MUST be sent to the address in the copy-
"The copy, if sent, MUST..." just as a picky wording change; otherwise, I
had to read it again to be sure of whether the e-mailed copy was a SHOULD
or a MUST.
> addr, and in the absence of that to the address(es) in the Reply-To
> header, and in the absence of that to the address(es) in the From
> header. The followup agent SHOULD also include a "Posted-And
> Mailed: yes" header ().
> [There is a suggestion that it should be allowed to include a Posted-
> And-Mailed in the body instead of that header. But I do not like
> specifying what body content should contain, and how would I interpret
> such a possibility if the body were a Mime multipart?]
Agreed. I'm not in favor of that suggestion.
> If this header is absent, a followup agent MUST NOT automatically
> email a copy of the followup article, but the user MAY do so at his
> discretion.
Perhaps add a parenthetical here "(such as by using a different command
than the default followup command)"? Or maybe not; not sure if the
clarification would be useful.
> 6.2. Posted-And-Mailed
> Posted-And-Mailed-content = "yes" / "no"
> This header, when used with the "yes" keyword, is used to indicate
> that the article has been both posted to the specified newgroups and
> emailed, either to the author of the article to which this is a
> followup (see the Mail-Copies-To header in section) or to the
> recipient(s) identified in any To and/or Cc header that may be
> present. The "no" keyword is included for the sake of completeness;
> it MAY be used to indicate the opposite state, but is redundant
> insofar as it only describes the default state when this header is
> absent.
If we're going to say that, then I'd rather see a Posted-To header that
lists the newsgroups to which the message was posted, since that provides
some additional useful information (and is already generated by some
clients, so there's existing practice). But I don't have any really
strong opinions about this whole area and have been fairly burned out on
arguing about it.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>