ietf-nntp NEWNEWS deprecation.

[Vincent Archer]

On Thu, Dec 04, 1997 at 08:35:46AM +0000, Clive D.W. Feather wrote:
> Vincent Archer said:
> > The answer is: why would NEWNEWS be more useful than an OVER/HDR?
> Efficiency.

True. And I have summarised in my message the exact circumstances in
which NEWNEWS is marginally more efficient than other fetching methods,
which are : if you are fetching a large amount of newsgroups and are
reading from a multiplicity of news sources.

> > is totally useless, because to READ news, you need a lot more than getting
> > articles.
> Not true. The economics of phone calls in this country mean that the best
> strategy is often to download all the articles in one fast session, and
> then do the processing afterwards. Or download the heads and then select
> the bodies you want for later.

I do agree. I was referring to the "watch-sized" reader presented as an
example of why NEWNEWS was useful, and proved that it was a flawed example,
because, if that watch-reader was too simple to use sophisticated protocol
queries, then it would also be unable to cope with the news it had downloaded
afterward.

But let's get back to phone calls economics. Which I now very well, since
they are similar in France :)

Tell me, if I am fetching news off demon, what would be faster for me?
Getting NEWNEWS over twenty newsgroups, receiving 1000 Message-Ids, then
asking 1000 (since I'm reading from demon, none of those Message-Ids are
already present here - they are all new) ARTICLE <some-large-id>? Or
doing twenty "GROUP x", intersped with 1000 "ARTICLE number". Oh yes, and
using the Xref provided in the headers to mark cross-posts in advance.

Pipeline of ARTICLE commands (in which I request the next article before the
previous has even begun to arrive) is broken only at the "GROUP x" switching,
and even there, I can always try to immediately request the first article
number in that group, and will check when I get the reply to GROUP if that
move was or was not useful.

> Pro-NEWNEWS:
>     It's in the RFC.
>     We find it useful.
> 
> Anti-NEWNEWS:
>     INN is crap.
>     Therefore we should get rid of NEWNEWS and force people to change.

Nope.
Anti-NEWNEWS:
	NEWNEWS usefulness is marginal
	NEWNEWS core functionality is duplicated elsewhere
	NEWNEWS is already disabled in a lot of real-world implementations

> The anti's just aren't addressing our issues: it's *IN* the RFC, and we
> find it useful. If INN had better code, would you be making this fuss about
> getting rid of NEWNEWS ?

You're right, probably not. I wouldn't use it either, because for all my
uses, it is totally useless, if not counter-productive. Just like I don't
care for 95% of the commands in M$ Word, except that it bloats the software
to intolerable levels.

I repeat myself: I don't want to erase all traces of NEWNEWS from the RFCs.
I just want that the fact that it is not used nor necessary in a lot of
real world cases is recognised. In other words, that the official documents
adhere to a reality in which no client can connect to a random server and
rely on getting NEWNEWS instead of an idealised view that would brand all
servers that have NEWNEWS disabled evil spawns of satan and pure crap that
should be eradicated from networld.

-- 
Vincent ARCHER  -=-=-  Herve Schauer Consultants -=-=-   archer at hsc.fr
Tel: +33 1 41 40 97 00                          Fax: +33 1 41 40 97 09
        01 41 40 97 00                                  01 41 40 97 09