|
Thanks for the great feedback! I received feedback from a couple colleagues. Below is my response to them. /Roger Scott wrote: Dr. Scott Law of Data #3: Hierarchical data structures are great, if it's
your hierarchy. Otherwise they suck. I like it! how do I know that <metabolism> is related to <picker>
instead of <Vineyard>? Great question! I am not sure. Perhaps that knowledge should be made available out-of-band (not hardcoded into the XML)? David wrote: the real issue is, what’s the data model separate from XML?
My thinking is that with most data-modeling problems many different models could be created. That is, there is no one, true model. For example, with my grape vineyard one expert modeler argues that the correct model is this: A Lot may have zero or more Pickers on it. This XML is a perfect representation of that model: <Vineyard> <Lot id="1"> <ripe-grapes>4</ripe-grapes> <Picker id="John"> <metabolism>2</metabolism> <grape-wealth>20</grape-wealth> </Picker> </Lot> <Lot id="2"> <ripe-grapes>3</ripe-grapes> </Lot> ... </Vineyard> Another expert modeler argues that the correct model is this: There are Lots. There are Pickers. A Picker may be located on one and only one Lot. This XML is a perfect representation of that model: <Vineyard> So which expert modeler has the correct view of the world? And thus, which XML representation is correct? It seems to me that both models/representations are well-suited to some consumers and horrible to others. So I am proposing this: Be model-agnostic in the XML representation. That is, give consumers a flat XML and let them parse it to represent whatever model is best-suited to them. Thank you! /Roger |