The one in tserver also counts across all deep copies of the iterator. It looks like the tserver one is used for providing compaction stats, and the one in core is used only for testing iterators (mainly to verify that seek optimizations are being performed correctly by checking the number of times next is called).
On Mon, Nov 25, 2013 at 8:08 AM, Josh Elser <[email protected]> wrote: > Sans the CountingIterator in o.a.a.t.Compactor incrementing an AtomicLong > every 1k, they are exactly the same class. > > > On 11/23/13, 7:55 PM, David Medinets wrote: > >> CountingIterator is declared in both of these places: >> >> org.apache.accumulo.core.iterators.system.CountingIterator >> org.apache.accumulo.tserver.Compactor >> >> The code looks a little bit different. Can one be eliminated? If not, can >> the names be changed to reflect the differences? >> >>
