2010/10/13 Paul Davis <[email protected]>: >> // This assumes keys is sorted (which I think is safe), but if not >> you can always sort them yourself > > This algorithm is what I was thinking of. Its basically the top N tags > example but just counting uniques. There is a gotcha in that you need > to sort the keys yourself because of a weird reduce thing I never > tracked down completely. IIRC, it was something about how the final > reduce runs that requires this. > > Otherwise, what Randall said.
It looks too complicated for my needs. I think I'm going to change my data model to avoid this problem. Thanks for your suggestions. Anand
