Page Layouts information is wrongly stored against a div element with a data-* attribute. This is invalid XHTML.

This is raised from a customer comment on CAC.

I believe the analysis is spot on but to summarise, we should strive to to keep storage format at close to XHTML as possible. Additional things we need should be in our namespace. The 'data*' attribute is not valid XHTML.

Likewise, these layouts render with the data attribute as well. I imagine this is done so we can recognise layouts when we copy and paste into the Editor. Confluence now renders output to HTML 5 so this is fine.

So regarding storage format. Could we put a custom namespaced attribute on the root div of the layout? Or better still could we create a custom element to represent the layout? Perhaps something like <ac:layout>. Obviously this should transform to a <div> when we render to view.

Anyway, the point here isn't to propose a solution, but instead just to call out that the implementation detail here is deviating from one of the goals of Confluence storage format.