From: Bruce Lilly (blilly@erols.com)
Date: Wed Jun 23 2004 - 07:53:50 CDT
Henry Spencer wrote:
> On Wed, 23 Jun 2004, Thorfinn wrote:
>
>>Can I request that the Chair make some sort of official declaration on
>>this topic? Otherwise we keep going around in circles, to absolutely no
>>point.
>
>
> Yes, I think we need a ruling on this one. The fundamental question is
> whether we are allowed, for news purposes, to add constraints on headers
> defined by the existing mail standard, or whether nothing at all can be
> said that restricts mail-defined headers further.
Henry,we've been over this many times before. RFCs 850 and 1036 adopted
the common (not "mail") message format defined in RFC 822. RFC 822 is
not under our jurisdiction (nor is MIME, which some WG members also
thought they could redefine). A message is a message, and all messages
need to be able to have common components (including the Subject field)
interpreted in a common way, otherwise established and widely-deployed
Standards Track protocols such as IMAP (which deal with messages, not
"mail" or "news" as such) would be adversely affected. And we don't have
jurisdiction over IMAP either, no matter how much some WG members keep
repeating the "IMAP must change" mantra. Aside from IMAP, there are
common UAs that handle news and mail, and such UAs use common code to
handle the common message format. That is a direct outcome of the choice
made over two decades ago when RFC 850 adopted use of the common message
format, discarding the earlier news-specific format; indeed the ability
to make use of common code was the impetus for that adoption, as clearly
stated in RFCs 850 and 1036. There are also existing gateways between
news and other applications using the common message format (mailing
lists being a prime example), which rely on the common message format;
i.e. they do not need to "translate" Subject, From, Date, Message-ID,
etc.; backwards compatibility -- prominently mentioned in our charter
(and still meaning that the standard we produce must not cause breakage
to existing protocols and infrastructure, no matter attempts to redefine
"backwards compatibility" otherwise)-- would be thrown out the window if
we were to redefine common components (if we had the jurisdiction to do so).
In short, you're asking if we have the authority to redefine RFCs 822
and 2822. The answer should be obvious. That's not in our charter, and
is managed by a separate IETF WG.
If a ruling from the Chair will stop these annoying and distracting attempts
to change standards that are outside of our jurisdiction, then I add my
voice to the call for a ruling.
> Historically, news has long imposed *many* extra constraints on headers,
> but our draft does attempt to relax a lot of them, partly in the name of
> easier mail compatibility.
Not just mail, compatibility with any other application of the common
message format (mail certainly being the most widely used). The changes
redress issues arising out of RFCs 850 and 1036, both of which were quite
clear about the adoption of the common message format and that the defining
RFC (822 at the time) had precedence in the case of conflicts. They are
also a result of the progress of Moore's law -- a number of the "constraints"
in 1036 were supposedly to permit simplified processing (and it should be
noted that RFC 2822 has also tightened the common message format syntax,
eliminating some of the broad flexibility that existed in 822, simplifying
parsing and generation of the common message format).
>>The fundamental question that seems to be unanswered is, "can we specify,
>>in this standard, the content of the Subject: header"?
>
>
> More precisely: "...specify, in this standard, for news purposes...".
> Nobody is claiming that we can change 2822's definition of it for mail;
> the claim is that we are the subject-area experts for news and it is
> properly our decision, not 2822's, as to whether specific 2822 rules
> apply to news.
That has also been asked and answered before. For two decades, news has
intentionally used the common message format. In that time, at least one
message access protocol has been developed and widely deployed, with
multiple implementations including deployment at large-scale installations.
That protocol (there may well be others in the same boat) has advanced
on the Standards Track and is based on the guarantee given in RFCs 850
and 1036 that RFC 822 is the defining document for that common message
format. That message access protocol is unable to differentiate "news",
"mail", or any other message in RFC 822/2822 format -- indeed there are
any number of instances where it is not possible to make such a distinction
(e.g a file in common message format, which might have been saved from
a "mail" application such as mailx, or by a "news" application such as
"trn"). Any attempt to change the common message format, whether as a
blanket change or "for news purposes", is going to affect a number of
protocols which are outside of this WG's jurisdiction and the knowledge
base of at least some of its "experts". That's *any* attempt, and
applies to attempts to change the syntax and/or semantics of Subject as
well as some other specific changes previously proposed. See for example
the following messages in the WG archive:
http://www.landfield.com/usefor/2003/Feb/0161.html
http://www.landfield.com/usefor/2003/Feb/0783.html
http://www.landfield.com/usefor/2003/Feb/0790.html
http://www.landfield.com/usefor/2003/Feb/0838.html
http://www.landfield.com/usefor/2003/Feb/0853.html
http://www.landfield.com/usefor/2003/Feb/0194.html
http://www.landfield.com/usefor/2003/Feb/0209.html
http://www.landfield.com/usefor/2003/Feb/0208.html
http://www.landfield.com/usefor/2003/Feb/0267.html
http://www.landfield.com/usefor/2003/Feb/0322.html
http://www.landfield.com/usefor/2003/Feb/0374.html
http://www.landfield.com/usefor/2003/Feb/0395.html
http://www.landfield.com/usefor/2003/Feb/0562.html
http://www.landfield.com/usefor/2003/Feb/0329.html
http://www.landfield.com/usefor/2003/Feb/0394.html
http://www.landfield.com/usefor/2003/Feb/0410.html
http://www.landfield.com/usefor/2003/Feb/0583.html
http://www.landfield.com/usefor/2003/Feb/0426.html
http://www.landfield.com/usefor/2003/Feb/0486.html
http://www.landfield.com/usefor/2003/Feb/0493.html
http://www.landfield.com/usefor/2003/Feb/0487.html
http://www.landfield.com/usefor/2003/Feb/0484.html
http://www.landfield.com/usefor/2003/Feb/0554.html
http://www.landfield.com/usefor/2003/Feb/0507.html
The "news format is not an instance of the message format" and "we can
make any damned changes to any standards we please" horses have long been
dead. Please stop flogging them.
> Oddly enough, the "unstructured" faction simultaneously points to 2822's
> definition of it as unstructured, and insists that we should ignore the
> explicit discussion of "Re:" in 2822 because that somehow isn't really
> part of the standard.
Please back up or retract your claim. Please supply supporting documentation
for the "insists that we should ignore ... in 2822" statement. I've been
following this discussion quite closely and cannot recall any such claim.
Indeed, I clearly recall just the opposite, viz. recommendation that the 2822
text be used by verbatim copying or by normative reference. I believe your
statement is a gross misrepresentation.