[NNTP] LIST EXTENSIONS (again)

[Russ Allbery]

Mark Crispin <MRC at CAC.Washington.EDU> writes:

> Support for 380 and 480 appeared in my code between April 14, 1994 and
> August 1, 1995; and apparently 380 came first.

I did some more hunting.  That hypothesis makes sense, given that the
reference NNTP implementation does have 380 in its nntp.h, but the
1.5.12.1 release doesn't use it anywhere, using only 480.  INN 1.0 used
only 480 in 1991, so presumably the change, if the guess is correct, was
made sometime before that.  If there was a change, it's not mentioned in
the changelog of the reference NNTP implementation, so it's hard to date.

That puts the date of a change somewhere between 1986, when RFC 977 was
published (which was right about the time that NNTP software was first
being deployed) and 1990, since INN has never had a mention of it at all.

My guess would be that you added the code for interoperability with a
fairly old C News site.  That would explain why I've never heard of it; at
the time it was still in use, I was using VMS and had never heard of the
Internet.  I don't expect there are many NNTP implementors who are still
writing code who remember the history.  Stan or Henry might remember.

> I respectfully point out that RFC 2980 was not issued until October
> 2000. Therefore, it can not be claimed that 380 was "broken" prior to
> that time.

Actually, RFC 2980 can't even declare things broken; it's not a standard.
RFC 2980 is just documentation.

In some sense, anything that anyone does with NNTP outside of the bounds
of RFC 977 can't be declared "wrong"; in practice, I'm pretty comfortable
calling things wrong if they don't behave like the reference NNTP
implementation and INN, until we get a new standard released.  That's the
best that we've had for quite a while now (which is, of course, why this
working group).

> Actually, 580 makes more sense than 480.  Ditto 481 and 482.  Someone
> evidentally misunderstands the meaning of 4xx response codes.  I've kept
> silent on this problem because NNTP is so broken in other respects that
> are more important to fix.  The world will continue to survive even if
> NNTP continues to misuse of 4xx.

4xx has never, from the very beginning of the NNTP protocol, meant the
same thing that it meant in other protocols.  So that one you can blame on
the original RFC 977 authors, but all subsequent status codes are just
staying consistent with RFC 977.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>