WWW NeXT client, telnet feature; Web traversal

Tim Berners-Lee (timbl)
Fri, 27 Mar 92 11:45:43 GMT+0100

Date: Fri, 27 Mar 92 11:45:43 GMT+0100
From: timbl (Tim Berners-Lee)
Message-Id: <9203271045.AA25109@ nxoc01.cern.ch >
To: Linda Murphy <murphy@dairp.upenn.edu>
Subject: WWW NeXT client, telnet feature; Web traversal
Cc: www-talk@nxoc01.cern.ch

> Date: Thu, 26 Mar 92 18:06:51 GMT-0500
> From: Linda Murphy <murphy@dairp.upenn.edu>

> My environment: a NeXT.
> ...I am running pre-release b of version 0.13 of the WWW application.
> I get ... Invalid access prefix for 'telnet://info.cern.ch'
> Can be one of news: but not 'telnet:'.

The NeXT client was frozen last summer, so it has no innovations which occured  
since then. I have it relatively high on my (rather large) agenda to bring it  
up-to-date from a prototype to  include the re-engineerd common library, and  
rerelease it.  (I was under pressure here not to put too much work into it because  
NeXTs were not official CERN workstations for a while: everyone wanted X. Now X is  
coming from other people, the pressure should ease.)

Another problem you will find is that the NeXT client can't cope with the very long  
identifiers returned by the latest WAIS servers such as the directory of servers.  
It just crashes becuase I put in an arbitrary hard limit (bad! :-().

Apart from telnet: it also can't handle gopher:.

I'm really sorry for that lack of functionality.  I use the app all the time myself  
so it bugs me too. I'll do it when I have time, but better server functionality I  
think should come first.  Someone in Hawaii whose mail address I don't seem to have  
(are you out there?) thought they might find six person-months to put into the NeXT  
app which would be great,

> Further down in the bug list, you mention serialisation and web
> traversal.  What do you mean by this term?

> --lam

I mean a feature to turn a web into a serial document, like to print it on paper,  
by traversing the web. This is really needed -- the world is looking for ways of  
turning text into hypertext, but the moment you do it, you want to turn it back  
again for people who want paper! A traversal, and concatenation followed by a sed  
file to turn it into TeX macros should cover it.

Also one imagines tools which traverse the web recursively in a breadth-first way  
looking for things -- interesting data and indexes for example. I image terminating  
the search as a function of number of links traversed modified by the  
"interestingness" of documents found on the way (judged by the words the contain  
matched against a query).  This is a step toward a "knowbot"-like tool for resource  
discovery. Now we have a real web to play with, we can start making such machines  
in earnest.