Re: news: and nntp: URLs

Tim Berners-Lee (timbl@www3.cern.ch)
Fri, 26 Nov 93 18:37:08 +0100

Date: Fri, 26 Nov 93 18:37:08 +0100
From: Tim Berners-Lee <timbl@www3.cern.ch>
Message-Id: <9311261737.AA00561@www3.cern.ch>
To: mitra@path.net (Mitra)
Subject: Re: news: and nntp: URLs

>From: mitra@path.net (Mitra)
>Date: Thu, 25 Nov 1993 09:37:57 PDT

>} other things. So you can't use these URLs in any sort
>} of reference except between people using the same
>} new server.
>
>Or as members of a list or search returned by Veronica for example.

Correct me if I am wrong, but Veronica uses the gopher
protocol, and returns gopher pointer, which can be converted
into gopher URLs. The news URLs which are returned are NOT nntp
URLs, but are pointers to a gopher gateway. Because a single
news gateway is used, there is no problem with people attempting
to dereference the article id with a different NNTP server.
This sort of gatewaying has nothing to do with news URLs.
It images the whole usenet world within one gopher server.
It doesn't scale, of course, but it can generate its selector
strings how it likes. This is not threatened by any URL spec.

>} 4. You can call it an access algorithm, but you have no way
>} of deriving this nntp: url from any other sort of name
>} such as the message-id esxcept by retrieving the message
>} [HEAD] using its message-id.
>
>This is where the biggest disagreement comes, I have two ways
currently
>of getting these URLs, they get returned by Veronica,

No they aren't -- see above.

> and they appear as
>the path in all the Gopher->News gateways.

They are not nws URLs here they are gopher URLs!
If you want to make a single big gateway somewhere
for all news articles, or even do it by group, you
will be abusing the way news works on the Internet.
News is a different special protocol which flooding
as it does is very efficent for certian types of
information. If you want to use gopher, use gopher.
It may be that if you run a large public access system
with a single point of access for all services, you
do not care too much about whether stuff comes by FTP
or NNTP. But on the Internet, the difference is very
importnat and one should not just use NNTP as FTP.
Or they guys wouldn't have written NNTP.

>Actually Tim, the argument went in the other direction.

>1) The news URL is not a URL, lets rewrite it so it is

This is what I meant bythe tomato example in my message to Peter.
The news protocol doesn't fit your URL/URN model. That doesn't
mean news is broken.

>If .newsrc was teh only place these things appeared, then I'd agree
with
>you, but I've named two other places group/article#'s exist and work
in IIIR
>appliations

Neither of which turned out (I think) to actually be using
pointers to NNTP hosts with group/article numbers. They
were gopher gateways which is quite different.

> (contrast this with zero places where I've seen applications
>that will actually retrieve a message-id style URL)

Hmmm. I could name you a dozen programs which do it.
But I have been told off for quoting existing practice :-)
Let's rest with the fact that it's better.

>- Mitra

Tim