On Mon, Dec 29, 2008 at 4:09 PM, Antony Blakey <[email protected]> wrote: > > And as far as writing is concerned, using name identity for values that > appear in multiple contexts makes it easier to write polymorphic code e.g. a > function that extracts the document id and revision from either a view row > or the document value (possibly) contained within. >
Which is exactly the thing I like about only having _id and _rev in documents. My being able to identify CouchDB documents by duck type is something I've come to enjoy in my coding. I think it makes for simpler applications in places. Splattering the world with meaningless "_id"s and "_rev"s only makes things more complicated. Polymorphic identity has it's place, but you really shoudn't have an _id / _rev pair on your hands, unless you've also got the document itself. Breaking that contract will produce muddled thinking about Couch's MVCC. -- Chris Anderson http://jchris.mfdz.com
