Push Button Paradise

Given all the bits and blood spilled over the recent XHTML Modularization Proposed Recommendation, I thought I'd share some background research on the topic.

This document seemed to jump from nowhere to the almost-final Proposed Recommendation state. Was this published in violation of W3C Process? It doesn't seem so. The official Process Document confirms: three separate entrance criteria are listed, all of which boil down to satisfying the Director, who himself announces the document to the W3C membership. If there was a process problem here, it came from the top. The move can be appealed.

But I don't think anyone will. XHTML Modularization is kind of, well, pointless. Perhaps as much as Infoset, it's from the shadowy realm of meta-specifications--it provides a base of terminology and structure for use by other specifications. It's weird to even talk in terms of software implementations of it.

So did this new version introduce new technical concepts? No, putting XML attributes in namespaces is pretty well established, and putting attributes in an XHTML namespace is just as possible now as it was before. Neither is it unprecedented. I'm not proud, but the XForms Recommendation shows an example that does give witness to the possibility of applying a namespace not-normally-on-attributes to attributes used in a foreign vocabulary. The concept is there, and has been a part of XHTML 2.0 for as long as it's included a reference to XForms.

Even if XHTML Modularization didn't have to go through another technical review, should it have? Good question. I think the overall W3C Process isn't well-equipped to deal with meta-specifications. Maybe meta-specifications themselves are the problem.

In summary, there's probably lots of more important things to get worked up over rather than this. Seriously, some of the comments have singed my industrial-strength snark filters. -m

posted at: 22:55 | under: 2006-02 | 1 comment(s)