W3C home > Mailing lists > Public > www-html@w3.org > March 2003

Re: Encoding of site structure ...

From: David Woolley <david@djwhome.demon.co.uk>
Date: Sat, 8 Mar 2003 20:09:17 +0000 (GMT)
Message-Id: <200303082009.h28K9Hq02290@djwhome.demon.co.uk>
To: www-html@w3.org

> the same meaning (there is no difference in HTML
> between a navigational hyperlink and a "normal" one).

I'm not entirely sure what you mean here, but rel and
rev attributes are allowed in a elements, so you can
sub-class links.

> What are these "external entities"?

The constructs used to include the named character entities into
HTML and XML; if you look at the specification, you will see that
these are actually separate files that get included into the 
DTD (even if they are actually hard coded in real world 
browsers.  As part of the document type definition, you can
specify the names of files that can then be included by using,
for example, "&navigationbar;".  In HTML you are not allowed to
do this in the <!DOCTYPE, but in general SGML and in XML you are
allowed to do so, but non-validating XML browsers aren't currently
required to expand the entities defined in this way.

> - When making a navigation bar from link elements,
> only those pages that are linked to directly from the
> displayed page are accessible. Displaying a complete
> navigation tree is impossible.

What I think you want is the document outline in PDF terms.  Again
this is the result of trying to create a closed compound document
with tools intended for loosely connected nodes.  HTML was designed
as a web language, not a compound document language.  However, I guess
one can create analogues of the various structures used by PDF in
the HTML world.

Incidentally, within a single, large node, proper use of Hn elements
allows the user agent to construct an outline tree; some of the minor
browser do this, an the html2ps tool used to create the PDF versions
of the HTML 4.xx specifications does it, meaning that the PDF version
has a proper outline tree without any extra coding.  I'm not sure if
Adobe's HTML to PS tool can do this; the examples I've seen seem to 
be at the file level, when it aggregates documents connected by link
elements.

I suspect one could construct a sideways view of a closed super-node
using an HTML document, and even improve its semantics by using
rel and rev on the links from that node, all using automated tools.

> - User agents _may_ display link elements for
> navigation. This means that the links has to be
> included in the document content too.

Sorry, but this argument contradicts the following paragraph - you
are saying that the browser cannot be trusted, even with style hints,
to display this information without it's being in the content part
of the page, whereas, in the following, you seem to be suggesting that
out of band data could be formatted,.

> And why does it have to be impossible for the authors
> to control the presentation of such navigation bars?
> Does the content of the pages _have_ to be the only
> thing for which authors may provide style sheets?
Received on Saturday, 8 March 2003 15:16:07 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:54 GMT