an answer I hope

Just looking at a credible wishlist for HTML, seeing how we're
doing.  This is from 
http://www.mcs.net/~jorn/html/net/wishlist.html
one of the best pages on hypertext style that I've seen:

> [Up] [Map] [No Prior] [Robot Wisdom home page]
> 
> A wishlist for HTML/Netscape/WWW
> 
>    * There should be two ways to select a link, one of which causes
>      you to go there, and the other to 'inspect' the nature of the
>      link-- where it leads, how much data it will transfer (w/ or
>      w/o gifs), etc.

REL/REV values, in HTML 2.1, kind of do this, since you can 
define yourself what the names of link types can be, beyond
the standard ones, and hopefully new standard ones will arise.
The spec was deliberately written not to care about what the
browser with the information, so write your own extensions etc.

The 'nature of the link' depends on your application, of course.
Legal cases laid out with links to 'jurisdiction', 'precedent',
'majority opinion', etc., or other things in other rhetorics.
We built a few such typed-link systems in the 80s for Xerox with
NoteCards, and much more recent stuff with SQL/SGML/HTML combos.

>    * Navigation commands need to be standardized and built into
>      HTML, so that we don't have to put our own 'up' 'next' 'home'
>      buttons on every page. (This will require supplying a map of
>      the topology, which the browser can translate into buttons.)

Done in the HTML 2.1 standard REL/REV values, which include
these and more.  Drawing a map or putting the buttons on the
toolbar is up to the browser, if it cares to do so.  It's done
with LINKs in the HEAD.  Browsers don't have to show LINKs as
they show Anchors.  More navigation will emerge.

NoteCards also had good map generation so I solved most of our
mapping problem in Interlisp-D in 1986, building on the good OO
components Xerox PARC had written.  Later when I beta'd the WWW
first generation client from CERN, I was disappointed that none
of the browsers supported typed links so I didn't bother to 
translate my stuff into HTML.  I guess I should have opened my
mouth more (sorry Tim).  Anyway it's there in 2.1.

>    * A 'virtual' anchor like "bigtext.html#%27" that would allow you
>      to link to a particular position (here, 27% down from the top)
>      in someone else's document, without them having to insert a
>      real anchor

Murray Maloney (murray@sco.com) or Liam Quinn (lee@sq.com) had an
idea like this that we discussed in a meeting once.  I don't know
if this was the problem we were solving but it seems like the 
same question:  reference to arbitrary points in another 
document.

This was/is a nice thing about Nelson's Xanadu, link to any part
of a page. 

>    * A sort of 'temporary bookmark' that you can create with a
>      single keystroke/mouseclick, for those occasions when you have
>      to choose between following an interesting link and continuing
>      to read an interesting page. It would make sure you get back to
>      the branch-not-chosen, eventually...

This is like tieing a ribbon around a particularly interesting
tree in the forest, like where you had your first kiss, and then
coming back to it when you are there another time.  Another of
the 'mapping issues'.  Mapping can't be done without typed links
or a LOT of heuristic/AI/rules/neural-net/markov-chain weirdness
on the contents of the page, which might include a human editor
and all the weirdness inside *that*.  Anyway...

Bookmarks are my personal pet peeve.  What I want is to have a
bunch of 'buckets', and be able to dump an interesting URL into
one or more of the 'buckets', without seeing any menus or forms
or 'bookmark editors'.  I want HTML pages generated by this that
I can use as jumping-off points for the various interests that
are represented by the buckets.  I don't want to hack Netscape
bookmark HTML files into good looking ones, I want this to be
automatic.
 
> I'm looking for where issues like these are discussed. This

Read html-wg@oclc.org for the standards issues, and 
www-talk for general discussions not about the HTML standard.
Warning, the traffic on the former will strain your SGML 
knowledge and people are not afraid to flame.  They're also
usually right, about SGML anyway. 

SGML compliance is important but the web can support a lot of
different formats.  I expect PDF, RTF, HTML, HTML Netscape 
version, SGML, Java, etc., to all survive alongside some much
weirder stuff.

> document
> from the WWW Consortium in Switzerland is the best I've found so
> far. But it's written in jargon, and very vague.

Heh.  No comment.  


> Here's another one that goes into a few more particulars
 
Webmaster stuff:  This is the new URL for the page you
referenced, please update your link so it points to W3 now:
http://www.w3.org/hypertext/WWW/DesignIssues/Topology.html
 
> [Up] [Map] [No Next] [Robot Wisdom home page] (Feedback)

If you email me about this, please use the address below as
passport.ca gets all my mailing list traffic and I don't clear
it out more than twice a week.
-- 
Craig Hubley                 Business that builds on knowledge
Craig Hubley & Associates    needs systems that build on the net
mailto:Craig@Hubley.COM    416-778-6136    416-778-1965 FAX

Received on Friday, 14 July 1995 04:11:34 UTC