To the WS-I, where I presented on behalf of the XML Schema Patterns for Databinding working group during the opening session of their plenary meetings to increase awareness of our issues and hopefully persuade more people to contribute patterns.
We've the WS-I above anyone else to thank for promoting the use of XML Schema as a way of literally describing SOAP payloads in WSDL, and the vendors who stack the board to thank for the widespread use of databinding tools which fail to consume XML Schema descriptions in oh so many different ways. So whilst they seem unable to help clear up the mess they created, it was a bit much to have the presentation derailed by heckling:
- Dr Evil:
- "Databinding is evil, so don't do it". Yeah, thanks! Whilst as a developer I may use XPath, it's my customers who insist on using your companies' sucky tools. What I'd like is to know how to publish schemas customers can consume, you know, interoperate with them.
- Sir Pedant:
- I made the mistake of mentioning how our charter covers versioning. "What is versioning, anyway?" Well I don't know precisely, but we've the good enough W3C TAG draft finding, XML Schema Versioning Use-cases and collateral from plenty of others who have dined out on Schema techniques for service evolution. Versioning is something any non trivial service has to consider. Let's just publish the commonly used patterns, and maybe toolkits will get around to not bailing at the first sight of
- Mr Relax:
- It was the assertion from one big-name vendor's rep that "using Relax NG is the answer" which rather took me aback. Great idea, I love Relax! Can we really ignore the history and expect the WS-I to deprecate Schema? Well no. It took a while to understand we should use Relax to model contents and then map to XML Schema. Unfortunately apart from James Clark's Trang, there isn't a standard mapping. Then Relax doesn't do "types", you still use XML Schema datatypes such as "xs:duration" which is where so many issues lie. Finally, schemas generated by Trang often don't work well with current tools in particular the ones touted by this bright-spark's company. It's a nice idea, but a bit like asking for directions and being told to start from someplace else.
So I was miffed that the message of our doing valuable, tangible, bug fixing work was lost in a melee of FUD.
Less than 24 hours later and I'm on a telcon where another standards rep from Mr Relax and Dr Evil's outfit relates difficulties mapping programming language data structures, such as an array, into XML. Sigh. How about a W3C working group to look into that very issue? Here's one we prepared earlier. Maybe you'd like to contribute?