From: Dan Kohn (dan@dankohn.com)
Date: Tue Apr 29 2003 - 19:12:53 CDT
In Section 3.2 of
<http://www.ietf.org/internet-drafts/draft-kohn-news-article-03.txt>, I
suggest dealing with this in the following way:
Following [RFC2822] syntax, the headers defined in this document do
not require a space between the ":" and the field's contents. (E.g.,
"Subject:Hello World" is acceptable, as opposed to requiring
"Subject: Hello World".) To be compliant with this specification,
news agents MUST support 0 or more spaces between the colon and the
field's contents. However, to maximize compatibility with the
installed base of news agents, implementers SHOULD use exactly one
space.
This would mean that INN was not compliant with this (new) spec, but the
spec gives implementers advice on how deal successfully with such
non-compliant software. This approach echoes the liberal/conservative
principle (Section 3.2 of RFC 791) and would, over time, encourage
convergence of news and mail software (to the degree it hasn't already
happened).
- dan
--
Dan Kohn <mailto:dan@dankohn.com>
<http://www.dankohn.com/> <tel:+1-650-327-2600>
-----Original Message-----
From: Henry Spencer [mailto:henry@spsystems.net]
Sent: Tuesday, April 29, 2003 13:47
To: Usefor Mailing List
Subject: Re: Differences between RFC 2822 and Usefor
On Tue, 29 Apr 2003, Bruce Lilly wrote:
> > As I think I've mentioned a few times before, INN rejects all
messages
> > that have headers without a space after the colon, and the upcoming
NNTP
> > standard revision will likely also require the space because that's
what
> > existing software expects and it makes the descriptions of various
header
> > retrieval commands much more straightforward.
>
> Sounds like a chicken-and-egg situation. INN isn't actually breaking
> on receipt of a message with the SP after colon.
No, it's just losing traffic, i.e. failing to interoperate. That is
more
than enough reason to (at best) relegate such a change to a "Future
Directions" section. Declaring a "flag day" on which all existing INN
systems must be upgraded would be ludicrous. This has to be accepted as
a
limitation of the existing infrastructure, and hence a constraint on the
standard.
> ...Let's not perpetuate a difference from 822/2822
> unless there's a very good reason to do so.
Preserving interoperability with a large installed base is usually
considered a very good reason.
(And no, interoperability with the installed base of mail systems is not
a significant issue by comparison: news-mail interchange has always
required non-trivial efforts by gateways, and there is no prospect of
eliminating that.)
Henry Spencer
henry@spsystems.net