RE: [xml-dev] "Introducing MicroXML, Part 1: Explore the basicprinciples

["Rushforth, Peter" <Peter.Rushforth@NRCan-RNCan.gc.ca>]

Hi John,

>I don't know which seven attributes you mean.  XLink 1.1 has ten
>attributes, but only one is required (xlink:href) for "simple conformance".

The seven attributes I suggested in the schema file I attached earlier (http://lists.xml.org/archives/xml-dev/201206/msg00073.html):

Specifically: 

xml:href, xml:src, xml:rel, xml:type, xml:hreflang, xml:method and xml:tref

which don't appear to have much in common with xlink.

XLink clearly pre-dates Fielding's thesis [1], which is the authoritative description of the architectural style of the Web (REST).  That authoritative description is a bit retroactive itself, but does itself stand on the shoulders of giants, for example Sir Tim Berners Lee's invention.  There are other giants involved even further back.

It seems possible or even likely that xlink was thought up without the benefit of a clear goal, or even if XLink's goal seemed clear, those goals were not informed by the REST paper.

[1] http://www.ics.uci.edu/~fielding/pubs/dissertation/top.htm

>The xml: namespace is not the place to put things that "might be nice".

All of these proposed attributes come from HTML.  The latter attribute I invented, as it seems to coincide well with RFC 6570, and as I mentioned to Uche, one shouldn't put invalid URIs in an href value, but once templates were processed, would be reasonable for a hypermedia application to perform appropriate actions on the resulting URI as described by the other hypermedia hints.

We use the things that the "giants" offer us to design our applications, on the assumption that they know what they are doing, and we should design our applications using those tools.  In the case of TBL and Fielding, those things just continue to grow in importance.  I am saying that those guys got links and hypermedia right, and we should put those things in their proper place, the XML namespace, particularly if we're rethinking XML in light of hindsight, as it seems to me MicroXML is an attempt.

While we seem to be on very different pages hypermedia-wise, you appear to have been convinced at one time that links belonged in the XML namespace - or perhaps I am missing some other context.

http://mailman.ic.ac.uk/pipermail/xml-dev/1998-June/004276.html : Cowan-XLink-Substantive-5.

Regards,
Peter Rushforth


________________________________________
From: John Cowan [cowan@ccil.org] on behalf of John Cowan [cowan@mercury.ccil.org]
Sent: June 24, 2012 6:44 PM
To: Rushforth, Peter
Cc: Uche Ogbuji; xml-dev@lists.xml.org
Subject: Re: [xml-dev] "Introducing MicroXML, Part 1: Explore the basic principles of MicroXML"

Rushforth, Peter scripsit:

> I'm not sure what the rule of least power would say about that, but
> from the KISS perspective, those seven attributes come from HTML,
> effectively, and if they were "burned into" XML, it would go a long
> way towards supporting the uniform interface in all implementations.


> Hence tref, which although does not come from HTML, might be nice
> for XML to put out there, since we have a newly created RFC for URI
> templates.

The xml: namespace is not the place to put things that "might be nice".

--
John Cowan                                cowan@ccil.org
At times of peril or dubitation,          http://www.ccil.org/~cowan
Perform swift circular ambulation,
With loud and high-pitched ululation.