🚀 go-pugleaf

[Dropping demon.service as this is getting rather technical ]

In article <31519680.42811766@news.demon.co.uk>,
Fred Hartley <fredkwh@incubus.demon.co.uk> wrote:
>
>Are you saying that an organisation wishing to implement an NNTP compatible
>news reader or to provide a news service using this protocol must do so by

NNTP does not have standards status, so there is no MUST.

>reference to a particular implementation and that any deviation from that
>implementation is "non-standard" ?

It's a little more fuzzy than that.  The reference implementation 
should be used where RFC 977 is ambiguous.
>
>Is this "reference implementation" fully documented and is that document in
>the public domain?

The only significant plain language document is RFC 977.  The
implementation is fully documented by the source code.  The source code
is copyright, but freely re-distributable, although there may be
restrictions on commercial sale or commercial derived works.
>
>Where may I find it ?

You should ask on news.software.b to determine exactly which version is
considered the true reference version.   The canonical site is
academ.com (possibly ftp.academ.com).  You probably also need C-News,
the canonical site for which is zoo.toronto.edu.

However, nntp and INN vie with each other for the latest extensions, and
most big sites (for good reasons) and a lot of leaf sites (often for bad
ones) use INN, so any client that won't work with INN, even though INN
actually violates RFC 977 in places, will have no market.

In particular, Netscape relies on behaviour in the reference
implementation, and INN, that is explicitly implementation dependent in
RFC 977.  I think that was probably carelessness, but it won't change.
My guess is that they implemented against INN, rather than thinking
about standards at all.

Generally clients should probe the capabilities of servers and fall back
to a core set of functions, and servers should not be broken by the
receipt of unknown commands.  A good client will work with only the
behaviour specified in RFC 977, plus any resolved ambiguities in the
reference implementation.

All of the above are also available, in some version, on any Linux
CD.

>
>Does this "reference implementation" contain any features which may only be
>used by others under a licensing agreement?

No.  Not unless you steal the actual source code.

>
>By whom is this "the most commonly held position" and have they publicly
>declared this view?

It is the position that you will get if you ask about the status of 
NNTP in the definitive (my judgement, I guess) newsgroups on this 
subject:  news.software.b and news.software.nntp.  These newsgroups
are monitored by the authors of the software which accounts for at
least 90% of the Unix and NT market (excluding Netscape), by my
estimation.  I suspect Netscape also monitor it, but they don't,
probably for commercial reasons, actively contribute; Netscape are
generally percieved to only be interested in standards for which
their software is the reference implementation.

>
>Has the appropriate body declared that RFC 977 is no longer the accepted
>reference document and that a more restrictive specification now applies?

RFC 977 has never been declared the accepted document - it has
information only status - it basically documents an old version of the
reference implementation, but with some ambiguities.  There is no formal
standard for NNTP.

You are living in a world where standards are set by consensus, rather
than a legislative process.

-- 
David Woolley, London, England          david@djwhome.demon.co.uk