RE: [xml-dev] xml:href, xml:rel and xml:type

[Liam R E Quin <liam@w3.org>]

On Mon, 2012-04-16 at 13:41 +0000, Rushforth, Peter wrote:

> 1. XML shall be straightforwardly usable over the Internet.
> 
> OK, this is not "mastery of the web", but the Web is an important
> application of the internet.

The goal was to put SGML on the Web. It was difficult to use SGML on the
Web at the time. The goal was not to replace HTML.

We never expected such widespread adoption. One feature of SGML remained
in the XML spec because we feared we might lose as many as a dozen
existing SGML users if we removed it.

Peter - the short answer is that painting the railway engine green won't
make more people travel by train. There's no single magic bullet. There
are barriers that could be lowered a little, but there also has to be
incentive, you have to offer things that people already know they want.

For trains, it's a perception of a frequent and fast service with no
need to book a seat.

For computer formats it's making something easier than what people
currently do - not equally easy, but easier. There might be some
traction in
<a http://example.org/>...</a>
for example, but that's not XML.

Grumpy pre-coffee responses follow... :-)

> Yes, others have pointed that out.  I'm thinking about RESTful use of
> XML.  REST is the basis on which the Web's protocol is designed.
> "typed" links are a fundamental part of REST, I think.

It's lucky that XPointer and XLink do not preclude such use then :-)
(and yes, I read Roy's thesis when it was published)

> So, my question is, what is the simplest change that could be made
> to XML which would enable its "success" on the web?

The simplest change is no change - XML is a success and met its goals.

If you mean, get XML used by people in preference to HTML...
* there's no direct way to get JavaScript used in plain XML
* the HTML DOM doesn't deal well with XML namespaces
* CSS doesn't deal well with XML namespaces
* XML cares about syntax errors and so do Web browsers; in principle
  Web browsers are not forbidden from correcting such errors as long as
  they don't claim the input was well-formed XML, but in practice the
  Web developer communities and the XML communities have remained very
  separate and the motivation to improve the XML experience in the
  Web browser is limited.  Namespaces in CSS became a recommendation
  last year, go figure.
 
> I wanted to hear from the technology implementers
> of what would break if such a change was made.

I don't see Web browsers changing for it - they don't support XLink
either.


> > If "mastery of the Web" _had_ been a goal, XML would have 
> > been very different, as Jeni Tennyson pointed out in her 
> > keynote at XML Prague this year.

I apologise to Jeni for the typo in "Tennison".

> Yes, I saw the video, but I forget this specific point.  Did she say
> how it would have been different?

  ``What might have been is an abstraction
    Remaining a perpetual possibility
    Only in a world of speculation.
    What might have been and what has been
    Point to one end, which is always present.''


> > To answer a more explicit question (from a personal 
> > perspective), the cost of introducing xml:href would be 
> > risking breaking existing XML systems,
> 
> I've tried putting these
> attributes on XML source and running XSLT against them, no
> obvious problems.

Names in the "xml" namespace are reserved, but the specs don't say
exactly what this means.

> I know there are vocabulary standards which use xlink.  Of course that
> means there are processors which use it too.

SVG is an example.

> > The biggest problem I see with XLink is that it did not 
> > address "link discovery through architectural forms" - the 
> > process of saying, "in this XML vocabulary, the "cf" 
> > attribute on any element with three vowels in the element 
> > name is a link constructed by using the template
> >   concat(@cf, "?", local-name(), "=", .)
> 
> I have been wondering about how to use URI templates in XML:
> http://www.rfc-editor.org/rfc/rfc6570.txt
> 
> Is this a similar idea?

Almost; here is a concrete example in that spirit:

<?link path="//person" name="." uri="http://example.org/people/$name"; ?>

You'd put this in the DTD (I know, Web browsers don't fetch DTDs because
of the FUD over "billion laughs", an attack also present in JavaScript,
and HTLM 5 deprecates processing instructions, but go with the flow
here...)

Now <person>daniel</person> would turn into
http://example.org/people/daniel

and a user agent could traverse the link based on some user action.

In a cooperative world :-) path would be any XPath 3 expression, but
only a few weeks ago I heard Web developers saying "XPath just reinvents
selectors and offers no new functionality, but unlike selectors cannot
be implemented efficiently" and with reasoning like that we've got a
long way to go.

I don't actually mean to be defeatist - it's a long road but a number of
people (including me) are trying to walk it, one step at a time, despite
the thorns in the path.

It took a decade for the Web developer world to adopt CSS. Now there's a
movement for people to be able to define their own elements in HTML. The
concepts are slowly coming.

Liam

-- 
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
Pictures from old books: http://fromoldbooks.org/