Re: Should encoding of site structure be standardized?

On Sat, 2003-03-01 at 21:09, Øystein Ingmar Skartsæterhagen wrote:
> 
>  --- Jared Warren <warren@cs.queensu.ca> skrev: > 
> > Your site navigation features are provided by RDF
> > Site Summary
> > <http://web.resource.org/rss/1.0/spec> which is
> > currently implemented in
> > Mozilla.
> 
> Can you give me some examples of sites using RSS for
> this purpose? There were only a few examples in the
> specification,

> and it seemed that defining "sites" was not the primary goal of RSS.

RSS's primary use on the Net is for syndicating news feeds. But whether
your pages change every day or not doesn't make a big difference in what
syntax is appropriate. Maybe rather than asking "what is this designed
for?" you should ask "does this fulfill my needs?"?

To use RSS for site maps I'd recommend using the content
<http://web.resource.org/rss/1.0/modules/content/> and/or threading
<http://web.resource.org/rss/1.0/modules/threading/> modules.

> Also: why isn't there a W3C recommendation for this?

RSS is a use of RDF, which *is* a recommendation. It's an application of
a W3C standard. (Think: RSS is to RDF as MathML is to XML?)

> Is RSS seen as a standard
> specification that UAs should implement, or is it just
> a proposal for one possible way of doing this?

If they implement RDF then RSS will work.

> What is actually RSS meant for? What I asked for, is a
> way of stating "this is a site", together with links
> to the pages in the site (sitemap)

It says "this is a set of related resources", which is as close to a
definition of "site" as I think you're going to get.

> and general site
> content (headings etc).
[deletia]

The Dublin Core module <http://web.resource.org/rss/1.0/modules/dc/>
does some of that, but you'd probably want a richer module for things
like header images. But of course: if it has a URI, it can be referenced
by RDF.

> > As for page elements like headers and
> > navigation bars, these
[deletia]
> > however doing it properly will require fully
> > processed generated styles
> > (eg: the content attrib) as are planned for CSS3.
> Would CSS be the properly way of doing this? As I have
> understood it, we can never rely on our CSS
> stylesheets being used by the UA, and I think these
> things (site headers etc) are so important to the
> pages that we can not leave them to some stylesheet
> that _might_ be used.

Why are they so important? Yes, it's a good user-interface guideline to
have site information on every content page, but the content is still
more important. Besides, if I'm using lynx or an aural browser, chances
are I won't be able to make much of your precious header.

> Althogh these things in any case seems to be a job
> which should be done by other languages than XHTML, I
> believe it should be stated in the XHTML specification
> how UAs are meant to retrieve and present such data,
> as it highly influences the way we write our XHTML
> pages.

Repeat after me: the Internet is not the Web. XHTML is just one of many
content formats to be found online. Just because it's been paramount in
the past, doesn't mean it should be for the future. Imagine: you sit
down to make a website, first thing you do is write up an RSS file, then
you start grabbing bits of content, in all sorts of different file
formats, and adding their URIs.

~ Jared Warren <warren@cs.queensu.ca>
Computing Science, Queen's University

Received on Sunday, 2 March 2003 00:49:45 UTC