Oh thanks.
Yeah, their docs state that it is internal. Kristina was telling me that my use case here would benefit from it, but I don't think it's exposed through node-mongodb-native. Which makes sense - the mongo docs specifically say that driver implementors should not expose it.
What I'm doing here might make more sense to handle at a driver level... I'm not sure. This will be the basis for a library I'm writing to sync the mongo state to the web browser. If other people find the oplog approach handy for that, it could implications for hosted DB solutions. My suggestion for hosting platforms would be to give everyone a personal oplog. You can do this by creating a capped collection under their access, and copying only events specific to their collections into their oplog.
Will be interesting to see if things go this way. I'm not sure how Meteor or Firebase are doing it, but I'd prefer a solution which is DB agnostic and open source.