Thanks for the post. I've spent the better part of the last year building an application based on XQuery, XHTML, and JavaScript with RESTful leanings. I imagine the comparable acronym to be XRJ (JRX doesn't come off very well).
From experience I agree with the observation that reducing data translation greatly increases the ROI of the development process. Although I continue to encounter quite a bit of ugliness with JavaScript, it plays an important role in my application and fits into the XQuery/XML architecture quite well. There is a lot of room for improvement of course.
Looking forward to digging into the proper XRX architecture as well.
Jason Monberg | June 5, 2008 07:20 PM
You write "Declarative techniques that use XML structures tend to accelerate the creation of domain-specific languages (DSLs)." What is interesting is that the vitality and fecundity of domain-specific markup languages has not, so far, translated into any vitality of domain-specific support languages. Quite the reverse.
Now I don't doubt that it is possible for such ecosystems to emerge (isn't that what the RDF project is attempting), but since a large portion of the XML value proposition is based on the use of COTS and general-purpose tools, I think they will be the exception rather than the rule.
I have two concrete examples. First, the use of Schematron (which operates direct on the markup) and W3C RuleML (when operating on RDF.) The second is that when the gap between the markup and the data model is great enough (e.g. which people variously consider as ugly markup or elegant model-driven development take your pick) then having some kind of domain-specific support (often just an API rather than a whole language) is a reasonable (but regretable?) approach.
Rick Jelliffe | June 10, 2008 04:59 AM
