RSS-Data: We already do it, sort of

http://radio.weblogs.com/0113297/2003/10/01.html#a237

I like the basic ideas behind Jeremy's RSS-Data proposal. I'd just like to add that we (and likely others) are already using RSS in this way. Greg at NewsGator described some of this in his recent case study of Triple Point's use of NewsGator.

Using RSS 2.0, which is namespace-enabled and extensible, we link and feed data from our source control system to our development teams, including project leadership and QA. Each "item" in the XML feed is a valid RSS "item", but also contains other elements in our namespaces so that purpose-built RSS readers can process a richer set of application-specific information using the same feed that our news reader, NewsGator, uses. So a single XML stream serves multiple masters. This is a capability that was not practically possible before XML.

Sometimes I think people forget that XML is both extensible and "un-extensible". There tends to be focus on creating rigid schemas to map this data or that process, while forgetting that it should be perfectly valid to attach your own elements to a vocabulary with the assumption that the parser that knows that vocabulary will safely ignore those new elements. Of course, many actual implementations don't work this way, but I argue they're broken and should be fixed. Parse what you know and ignore the rest. Otherwise, XML is just another implementation of statically-typed C structures, C++ objects or SQL tables, and adds no real value. The beauty and power of XML is in the flexibility, especially on the reading and parsing end of the equation.