RE: Arch Doc: 18 Sep: abstract

> -----Original Message-----
> From: Roy T. Fielding [mailto:fielding@apache.org]
> Sent: Tuesday, September 23, 2003 4:53 PM
> To: David Orchard
> Cc: noah_mendelsohn@us.ibm.com; 'Ian B. Jacobs'; www-tag@w3.org
> Subject: Re: Arch Doc: 18 Sep: abstract
>
>
> > What do you mean by "active" versus "descriptive" links?
>
> Active links are actuated by processes that traverse the information
> with a hypertext engine.  Anchor href, img src, and stylesheets fall
> into this category.  Descriptive links are supplementary information,
> such as xmlns or <link>, that establish a relationship between
> resources without an expectation that it will be actuated during
> normal processing.  Those links might be the focus of other actions,
> such as SemWeb stuff, but not hypertext.
>

Gotcha.  So "active" roughly means that there is an expectation of
dereferencing the uri.  Should we define these terms - link, active,
descriptive - later in the document, such as the glossary?  I think so.  I'd
like to use the existing terms, such as dereference and URI.  Something
like:
link: a component containing a URI and other data, ie attributes or elements
containing URIs.
Active link: a link containing an active identifier.
Active URI: a URI that may be dereferenced
Descriptive link: a link containing a URI that is not intended for
dereferencing
Descriptive URI: a URI that is not intended for dereferencing.
hypertext: an instance of a language that is intended to be rendered and
contains active links.

I'm obviously just trying to get these terms consistent throughout our
document.

> > I have kind of thought that "hyper" is associated with
> generally human
> > targetted.  XLink, intended for hyperlinking, has "show"
> and "actuate"
> > axis
> > because of the "hyper" part.
>
> Er, I don't know about that (MOMspider is not human, so I think that
> distinction has never held true).

Hence why I used the term "generally".  I'm vewy vewy vewy caweful and qwiet
when hunting wefinitions.  Surely though, the reason that MOMspider exists
is to facilitate human interactions?  That is, a human wants to find a page
with some particular text, and MOMspider has indexed such pages.  The use
case of where hypertext and MOMspider exist and no human is involved isn't
very compelling compared to the human-centric use cases.  Thus hypertext
seems to be intended for human consumption, though not restricted just human
consumption.  I contrast this with something like WSDL definitions, where
it's generally meant for software to read and write it.

Would you consider WSDL definitions or RDF files that are "on the web" to be
hypertext?  I don't think you do, because you talk about a hypertext engine
earlier and I don't think you think of my company's WSDL tools as a
hypertext engines.  And you then mention a "UI" style being named hypertext.

Where I'm going is that I think the web is about more than hypertext (a set
of representation types) and seems to me to be more about the agreement to
use standardized identifier syntax in the form of URIs in those
representations.  Thus the "web" is simply the complete set of active URIs
and dereferenced representations.  A corrollary is that a given resource is
"on the web" if it has an active URI.  At least I think that definition of
web, on the web, etc make a consistent set.  I'm not sure whether describing
the web as in terms of just active URIs, or resources that have active URIs
makes much difference.

A couple more definitions into the mix...
Web: the set of active URIs and dereferenced representations
"on the web": used for resources that have an active URI.
hypertext web: a subset of the web where the representations have hypertext
types.

> "hyper" is a synonym for "extremely
> active", and the reason the UI style was named "hypertext" is because
> the rendered text (whether it be bytes, image, or stream) is active in
> the sense of being the control-point for actions.  Thus, hypertext
> is equivalent to "activetext", where "text" refers to a rendered set
> of information.
>

I must admit that I find calling rendered images "text" counter-intuitive.
I would have thought that hypermedia seems a little more accurate.

I will separately post my suggestions for the abstract.

Cheers,
Dave

Received on Friday, 26 September 2003 01:16:25 UTC