fastutil ConcurrentMap

trafficstars

Is it on the roadmap to add ConcurrentMap implementations? It would be nice to have an atomic putIfAbsent() (something less heavyweight than SynchronizedMap).

Aug 15 '17 18:08 dmolesUC

Well, putIfAbsent has been already implemented and if you synchronize a fastutil map it will be atomic, too...

On August 15, 2017 8:14:00 PM GMT+02:00, David Moles [email protected] wrote:

Is it on the roadmap to add ConcurrentMap implementations? It would be nice to have an atomic putIfAbsent().

-- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/vigna/fastutil/issues/81

-- Sent from my Android device with K-9 Mail. Please excuse my brevity.

Aug 15 '17 21:08 vigna

The difference between fully synchronizing the map and having a concurrent map is (as far as I know) that the synchronized map usually gives a stronger atomicity than required. A concurrent hash map for example might lock each bucket individually or use read-write-locks compared to always locking the whole map.

Bottom line: Definitely possible to do, but a lot of work (testing atomicity is a real pain ...). One definitely could "fastutilize" the ConcurrentMap interface (this shouldn't be too difficult) and then do the same for the reference implementations.