W3C home > Mailing lists > Public > uri@w3.org > August 2004

Re: Fwd: draft-hoffman-news-nntp-uri-00.txt

From: Charles Lindsey <chl@clerew.man.ac.uk>
Date: Fri, 27 Aug 2004 18:58:42 +0100
To: uri@w3.org
Message-ID: <opsdej74e46hl8nm@clerew.man.ac.uk>

On Fri, 27 Aug 2004 11:22:25 -0400, Al Gilman <Alfred.S.Gilman@IEEE.org>  
wrote:

> At 7:20 PM +0100 8/26/04, Charles Lindsey wrote:

>> I have now proposed some syntax.
>
> Yes, thank you.
>> When we are agreed on that, then some semantics could follow.
>
> [Not sure I would think we can 'agree on' syntax without agreeing on  
> semantics.
> My religion says model first, then bind to syntax.  Not sure it matters.]

Well Paul's syntax was so badly broken that it needed to be mended first  
:-(.

Anyway, the syntax at least documents what features we think we are trying  
to implement.
>
> Is this what we are after?:
>
>    Requirement: describes actual implemented practice
>
>    Optimization factor: within the range of what is possible in that  
> regard,
> comes as close to being on a track of cleanup and rationalization of Web
> usage as we can.

More or less.

> The evidence that I have gathered so far is thin, but it does provide a  
> case of sorts for the
> following ideas:
>
> - Newsgroups are used for product support and for community-of-interest  
> caucuses in intranets.

Newsgroups are primarily used for the global Usenet. Intranets and private  
groups are relatively small compared with that, though they may turn out  
to be the main users of the nntp URL.
>
> - Accessing newsgroups with a newsreader is more usable under screen  
> reader conditions than through some web gateway, unless the gateway has  
> been rather carefully constructed.  DejaNews is not bad but the sponsors  
> or builders of interanets don't want to be beholden to that outside  
> interest and the news: URI that drops the reference into a newsreader  
> that is already in the browser on the desktop is a simpler solution not  
> involving as many strings attached.

Generally, web interfaces to news are a Bad Thing. If you can see it in a  
genuine newsreader, then you are in a position to generate followups and  
look around for related articles in the same group, and so on. OTOH,  
Google searches tend to throw up news articles that were posted years ago  
and long expired from news servers, so in that case the Google interface  
is all you have.
>
> So there's a case for giving a news: URI as an alternate to an http: URI  
> where the caucus or message
> is cited in Web content (as an access enhancing technique).

I think you are more likely to find these URLs (or at least ones using  
<message-id>s) in the body of other news articles, as a form of  
cross-referencing. Or you will find them in email messages.

OTOH, URLs referring to a whole group are more likely to be found on web  
pages, drawing your attention to the newsgroup as a whole.

> My thought had been to put nntp: URIs on a phase-out path.  Nowadays I  
> would be delighted to get an IETF document which provides a best-fit  
> description of what actually interoperates.
>
> Remind me: are <article-number>s portable across servers?   Or are these  
> independently assigned
> as the messages are received at the server?

The latter. Which is why the <server> is not optional in the nntp URL. So  
it only makes sense for communication between people who have access to  
the same server (which is mostly on intranets).

But, interestingly, there was a suggestion on one of the opera.*  
newsgroups today that someone whose locally stored newsgroups had got  
corrupted in some way should use an nntp URL to force them to be retrieved  
 from the server again. So there is a genuine use for you.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl@clerew.man.ac.uk      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5
Received on Friday, 27 August 2004 20:44:40 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:08 UTC