Skip to content

Release 0.9.0
Compare
Choose a tag to compare
@dweiss dweiss released this 08 Jun 11:32
· 59 commits to master since this release
0.9.0
fab14fd

This is the API-breaking release of HPPC in a longer while. A new interesting associative container has been added (see worm hashing issue below) and some changes have been made to get rid of the Scatter*-type containers. You can now safely use either worm or regular version of hash maps and sets.

Resolved issues:
https://github.com/carrotsearch/hppc/milestone/1?closed=1

JavaDoc:
http://carrotsearch.github.io/hppc/releases/0.9.0/api/

New features and API changes

  • Java 11 is now required to compile HPPC. The resulting binary JAR is Java 1.8 compatible (Dawid Weiss).

  • GH-24: Support indexRemove in KTypeVTypeHashMap,KTypeHashSet,KTypeWormSet (Bruno Roustant).

  • GH-20: KeysContainer of WormMap is not public (Haoyu Zhai, Bruno Roustant).

  • GH-13: Add automatic module name to the generated JAR's manifest ("com.carrotsearch.hppc").

  • HPPC-179: Update java template parser to support Java 8.

  • HPPC-186: A different strategy has been implemented for collision avalanche avoidance. This results in removal of Scatter* maps and sets and their unification with their Hash* counterparts. This change should not affect any existing code unless it relied on static, specific ordering of keys. A side effect of this change is that key/value enumerators will return a different ordering of their container's values on each invocation. If your code relies on the order of values in associative arrays, it must order them after they are retrieved. (Bruno Roustant).

  • HPPC-176: A new set of associative containers implementing Worm Hashing has been added. This strategy is appropriate for a medium sized maps and sets (less than 2M entries). It takes more time to put entries in the map because it maintains chains of entries having the same hash. Then the lookup speed is fast even if the map is heavy loaded or hashes are clustered. On average it takes slightly less memory than KTypeVTypeHashMap: even though it allocates more data structures, the reasonable load factor is higher (it varies around 80%) so containers enlarge later. (Bruno Roustant, Aleksandr Danilin).

Improvements

  • HPPC-191: Improve Accountable implementation (Haoyu Zhai)

  • HPPC-183: Simplify IndirectSort comparator to use IntBinaryOperator.

  • HPPC-177: Modernize the build system to gradle and make it work with IntelliJ.

  • HPPC-184: Use closures where possible to make the resulting JAR smaller.

Bugs

  • HPPC-187: ObjectIdentityHashSet redistributes keys according to key.hashCode rather than object identity's hash code. (Bruno Roustant).
Assets 2
Loading