Re: ***SPAM*** [xml-dev] Resource relationships

["Liam R. E. Quin" <liam@fromoldbooks.org>]

On Sun, 2020-06-14 at 20:31 +0000, Hans-Juergen Rennau wrote:
>  Oh yes, I emphatically agree, these were key points missing:* XPath-
> based URI construction
> * XPath-based identification of the link context
> One more question: WHERE should those link definitions, as you
> sketched them, be stored

In a separate document, rather like an XML Schema or DTD.

> Suggestion, Liam: if this topic is still interesting to you, you
> might check out a related work from the JSON domain, taking care of
> URI construction (via URI templates + JSON pointer) and the selection
> of link context:
> https://json-schema.org/draft/2019-09/json-schema-hypermedia.html

Thank you - for XML, sadly, i think it too late.

> Cheers,Hans
> PS: Unfortunately, I did not understand your remark: "Combining this
> with mapping elements to HTML equivalents for search
> engine purposes, and maybe with automatic/external namespaces that i
> proposed at one point"

You don't tend to see raw XML documents much on the Web. After all,
search engines don't know how to index it usefully, and don't know how
to display the results. So an "architectural forms" approach that says,
in this voacabulary, para behaves like html:p and title behaves like
html:h1, would enable search engines to label snippets with document
titles and to know that a phrase would not* continue across para
boundaries. Today this could also be done with CSS.

The automatic namespace proposal was a simple XML file, itself with no
ise of namespaces, that could say things like, "an element with local
name svg appearing in the HTML namespace is to be marked as being in
the SCG namespace instead, along with its children", enabling mixing of
 namespaces to "just work" in most cases with no colons.

It was scuppered by a counter-proposal from some XML people with a
version of such a document that _did_ use namespaces on everything,
which made the browser people run screaming.

Liam

> 
>     Am Sonntag, 14. Juni 2020, 16:04:14 MESZ hat Liam R. E. Quin <
> liam@fromoldbooks.org> Folgendes geschrieben:  
>  
>  On Sat, 2020-06-13 at 20:24 +0000, Hans-Juergen Rennau wrote:
> >   Liam, please help me make sure that I understand you correctly,
> > when
> > you say:"What was needed was something like HyTime Architectural
> > Forms, to say,
> > in this DTD, in this document, such-and-such an attribute, or
> > element,
> > or combination, represents a link, and _this_is how you construct
> > the
> > URL from it."Do you mean *URI construction* from instance data? Do
> > you want to point out this difference: whilst XLink presupposes
> > ready-to-use URIs contained in attributes, it fails to address the
> > construction of URIs from document data, e.g. in the style of URI
> > templates (RFC 6570)?
> 
> I mean two things.
> 
> First, one should have been able to have said, in this vocabulary,
> that
> some element's content is connected to other documents as a link;
> second, one should have been able to have constructed an IRI (they
> were
> URLs back then) in the manner of URI templates, yes.
> 
> Consider potential explicitly-marked-up links in the following:
> 
>   <p>When <place>King’s College</place> was founded, the circle had
> not
>   yet been invented, which is why <person>John Wood</person> wrote
> his
>   seminal <work><title>Rectangles and Quadrangles</title></work>.</p>
> 
> One might have,
>   <link match="place" method="get" uri="lower-case(translate(' ', '-
> ', 
> .))" />
>   <link match="work[title]" uri="/search">
>     <with-param name="title" select="title" />
>   </link>
> (with missing components such as method, server, port etc. having
> sensible defaults based on context).
> 
> Combining this with mapping elements to HTML equivalents for search
> engine purposes, and maybe with automatic/external namespaces that i
> proposed at one point, would be a very different approach than XML
> Schema :) and would be more focused on meeting needs of people who
> use
> Web browsers and Web search.
> 
> Around the time XLink was being developed, though, there were still
> SGML people saying the Web wouldn't last, wouldn't scale, would go
> away
> soon.
> 
-- 
Liam Quin, https://www.delightfulcomputing.com/
Available for XML/Document/Information Architecture/XSLT/
XSL/XQuery/Web/Text Processing/A11Y training, work & consulting.
Barefoot Web-slave, antique illustrations:  http://www.fromoldbooks.org