Embedded XHTML is tricky?

When I first downloaded the XHTML 1.0 schemas, and realized that they were all completely separate, with no mutual reuse, I thought it was some type of lunacy. Of course, I hadn't considered at that point that all three subsets had to exist within the same namespace. It just appeared to me that it would be unnecessarily difficult to maintain.

Later, when I tried to include two of the schemas into my XML document, casually attempting define the same namespace twice, I realized that this was just not going to work, and then I understood why the schemas may have been separated like this.

It makes sense that there is only one namespace, since the union of all of these schemas is valid for that namespace. But at the same time, there is a desire to define separate schemas so that authors can guarantee that they are working within the subset they wish to target. The problem is of course, that XML Schema just isn't up to the challenge, and the present solution is a kludge. And, as is invariably the way with kludges, they tend to crop up again as new problems down the line. So with these schemas, we are unable to define XML vocabularies that stores multiple XHTML documents using different XHTML subsets.

I believe that the WAI group had this same type of problem when they created their own subset of HTML that would meet the Accessibility Guidelines. They used Schematron to solve that problem since it has support for co-occurrence constraints, and can easily be used to tighten an existing schema. All things being equal, this is the *right* way to do it, and would also allow authors to validate that a document is a valid XHTML instance, regardless of the XHTML subset it uses (another problem with the current approach).

Of course I am not for a moment suggesting that this ought actually happen. I am just drawing the groups attention to the other consequence of the current design, in case they are not already aware of them. I apologise in advance if I appear rude, as this is certainly not my intention. But isn't it always preferable to know the complete picture, if it is not already known.


Best Wishes, Dominic.