I stumbled over a tweet from Mike Amundsen where he essentially asked people to name some more “widely-used hypermedia-types” beside (X)HTML and Atom. Turns our Mike collected the findings and made it available at http://amundsen.com/hypermedia/. Cool. Thanks!
Couple of days later I read Linking data in XML and HATEOAS where Wilhelm contemplates about Linked Data etc. The last sentence of his post reads:
Anyone know why XLink was abandoned, or why linked data doesn’t follow this concept?
My hunch is that XLink didn’t have the expected uptake and hence failed to serve as a basis for a light-weight and simple way to link data on the Web.
As I’ve argued in a previous post, typed links are essential for true HATEOS, however I wonder if we’ve only scratched the surface of this …
I use XLink internally every day. I also use it in custom desktop and console clients that use the HTTP protocol.
And I also think we have lots of opportunity to explore hypermedia links in new and existing media-types.
Posted by Mike Amundsen | 2010-02-10, 18:43Mike,
Thanks a lot for your comment – very interesting to learn this. Is there anything publicly available I could have a look at? source code-wise, documentation, etc …
Cheers,
Michael
Posted by woddiscovery | 2010-02-11, 09:56XLink failed because of a mix of issues. its model was basically undefined, it only was a syntax. the syntax is not all that great and a lot of people disliked it because it was namespace-heavy. the vaguely defined model was very much centered around UIs for links, i.e. behavior of links for users interacting with them. the XLink designers had a very strong background in document-centric hypermedia/hypertext, but zero background in machine-oriented “hyperdata” or REST. XLink is used internally and survived in some specs in a limited form, but other than that, it was a near-100% failure. and the most promising idea of XLink, the idea of a “linkbase”, a place where you could go to find links, was completely underspecified, and this is what some of the “discovery” work for SemWeb technologies tries to do in a better way today.
Posted by dret | 2010-02-10, 19:02Erik,
Thanks for your detailed background story, much appreciated! Question now for me would be: what are the lessons learned. Can we somehow benefit from XLink concepts? What pieces would you advise me to look at?
Cheers,
Michael
Posted by woddiscovery | 2010-02-11, 09:58I thought it was dead seven years ago and explained my theories why in an XML.com article at the time at http://www.xml.com/pub/a/2002/03/13/xlink.html .
Bob
Posted by Bob DuCharme | 2010-02-10, 19:32Michael:
I don’t any X:Link example to show, but here is a link to a live sample app that uses the related X:Include (which I find very useful).
The link also points to a short article on the sample along with the available source code.
Posted by Mike Amundsen | 2010-02-11, 15:00… ahm, can it be that you’ve forgot to include the link, Mike?
Posted by woddiscovery | 2010-02-11, 15:03Oops!
http://exyus.com/xcs/server-mashup/
http://exyus.com/articles/server-mashup/
Posted by Mike Amundsen | 2010-02-11, 15:04Thanks, Mike – looks interesting, though I have to dig deeper before I can ask good questions, I guess
Posted by woddiscovery | 2010-02-11, 15:20Michael:
No magic here. I leverage the X:Include to do “mashups” that might otherwise be done via scripting (XmlHttpRequest, etc.). In this sample server script (“Controller”) treats the XML document as the “Model” and the XSLT document as the “View.”
Posted by Mike Amundsen | 2010-02-11, 15:25