RE: [xml-dev] Xlink Isn't Dead

[peter murray-rust <pm286@cam.ac.uk>]

Back in the days (ca. 1997) when XML was bright and shiny it had 
three components:
- XML Core (I think this was the term)
- XML Style (now XSL-*)
- XML Link

I was optimistic and naive enough to believe that XLink was going to 
happen in the same way as the first two (which IMO are great 
successes). I assumed that it would acquire toolkits like the first 
two. I personally hacked quite a lot of code to support XLink at a 
prototype level. There are even examples in CML. I have been very 
disappointed that it hasn't really happened.

I need XLink for semantic relationships. I assumed that these were 
implied in it as well as "rendering". From the recent discussion it 
seems semantics is in a minority. When everyone talks about the 
S/semantic W/web
we are still obsessed with sighted humans.

As a result I have had to implement my own linking structure in CML. 
(Yes, I also use RDF, but it isn't cuddly). I need a strongly typed 
bidirectional link (i.e. the link knows what the type of the element 
is at the end). I want to point from an <atom> to a <molecule> and 
know that the link will check the target is of the correct type.

So I waited for an XLink toolkit. (It didn't have to be bloated like 
so many of the modern XML specs and tools.) None appeared. So I have 
had to do it all myself - the spec, the examples, the semantics, the 
code. What a waste of my time to end up with a system incompatible 
with anything else.

P.


Peter Murray-Rust
Unilever Centre for Molecular Sciences Informatics
University of Cambridge,
Lensfield Road,  Cambridge CB2 1EW, UK
+44-1223-763069