forwarded... [Explicit Linking is Impossible]

Tim Berners-Lee <timbl@www3.cern.ch>
Date: Thu, 6 May 93 18:59:40 +0100
From: Tim Berners-Lee <timbl@www3.cern.ch>
Message-id: <9305061759.AA18296@www3.cern.ch>
To: aronsson@lysator.liu.se
Subject: forwarded... [Explicit Linking is Impossible]
Cc: www-talk@nxoc01.cern.ch
Reply-To: timbl@nxoc01.cern.ch


> Date: Wed, 5 May 1993 23:24:26 +0200
> From: aronsson@lysator.liu.se

> [..]  If the destination document is maintained by someone else, he
> should be notified that I have a link going there, so he doesn't
> rename or reorganize his document without notifying me again.

In HTTP there is another SBNIBAYAFAIK feature that the client can
quote the "referee" (ie link source) when following a link. The
features is designed to allow backward links to be built up as the
forward links are followed.

> Date: Thu, 6 May 1993 01:21:40 +0200
> From: aronsson@lysator.liu.se

 

> What exactly does "pre-load a search" mean, and how do I implement
> this in an HTML document?

If there exists an index which covers the area (like if I got around
to WAIS-idnexing all of that, you would quote something like

	http://info.cern.ch/hypertext?style

and point the reader at the results of a wais serach for style in
the database.

> What if Tim and his WWW service moves from CERN to a new place?   
What
> if Tim decides to rename "WWW/Provider/Style" to "WWWSTYLE.DIR"?

	A big difference between the http deamon and the ftp daemon
	is the document name to filename mapping. Though I wish
	I had named the documents "/InformationSystems..." rather
	than "/hypertext", I won't change the document names even
	if I move the files. I will write a server which maps the  
document names into their new places. In fact, already you are
	spared making references to
	"/Net/dxcern/userd/timbl/hypertext..." thanks to the
	mapping.  Look at the mapping as a local name server so that
	within the server the path is in fact a name not an address.

	[Of course filenames aren't addresses either. I wouldn't like
	to have to count how many mappings occur before we get to
	the disk track/sector/cylinder numbers.]

> If someone can pull this rug, why should I stand on it?


	If I want to, I cn turn the server off. But I don't want to.
	It is in my interests to maintain the pointers.
	
> I would be happier to refer to an ISBN number, because then I only
> specify the publisher, and not the library where I found the book,  
and
> there would be an organization for registering new ISBN numbers  
that
> would stop an author from changing a documents name.

Have you joined uri@bunyip.com to talk about naming and addressing
of document resources?  If not, you might like to. Mail uri-request  
of course not the list.

Tim