Here's an idea I had for establishing new node types as extensions to OPML 2: All new attributes and elements introduced beyond the OPML 2 spec must belong to a new XML namespace. So, what if the XML namespace URI was also the value used for the "type" attribute on nodes? I could see all sorts of wrangling going on coming up with registries and conventions for values of the "type" attribute, but URI are good unique identifiers. So, why not make them serve for both the namespace and the node type?

Good morning, all. Is rainy outside, but 54 degrees F. This seems like a good compromise to me.