v0.24.0

New Features:

• Implemented equals and hashCode methods for hasher objects
• Introduced serialization and deserialization methods for hash stream states (getState and setState), enabling snapshotting of intermediate states and resuming processing from saved points. Additionally, hasher objects now include a hashStreamFromState method for initializing streams from a saved state.

Bug Fixes:

• Resolved issues with the equals method for hash streams introduced in version 0.23.0.
• Addressed edge case issues affecting hash streams when processing extremely large data inputs (≥ 2⁶¹ bytes).

Full Changelog: v0.23.1...v0.24.0

v0.23.1

Bug Fix:

• Fixed issue when creating packed arrays with bit sizes >= 2^31 - 7 = 2147483641 (corresponds to roughly 256MB). For example, for a packed array with a component size of 64 bits, this limit was reached for an array size of 2^25 = 33554432.

Full Changelog: v0.23.0...v0.23.1

v0.23.0

New Features:

• Added getHasher method to hash streams to retrieve the underying hasher instance
• Added new methods to hashers to hash integers directly:
  • hashIntIntToInt
  • hashIntToInt
  • hashLongToInt
  • hashIntIntToLong
  • hashIntToLong
  • hashLongToLong
• Implemented equals and hashCode for hash streams

Full Changelog: v0.22.0...v0.23.0

v0.22.0

New Features:

• Added the ConsistentBucketSetHasher interface for implementations that support bucket removals in arbitrary order
• Added the ConsistentHashing.jumpBackAnchorHash factory method for creating a ConsistentBucketSetHasher based on JumpBackHash and ideas from AnchorHash and MementoHash

Full Changelog: v0.21.0...v0.22.0

v0.21.0

New Features:

• added convenience functions for mixing/hashing integers and longs to an int
  • Hasher32.hashLongLongToInt
  • Hasher32.hashLongLongLongToInt
  • Hasher32.hashLongIntToInt
  • Hasher32.hashIntLongToInt
  • Hasher32.hashIntIntIntToInt
• added convenience functions for mixing/hashing integers and longs to a long
  • Hasher64.hashIntLongToLong
  • Hasher64.hashIntIntIntToLong
  • Hasher64.hashLongIntToLong

Full Changelog: v0.20.0...v0.21.0

v0.20.0

New Features:

• added XXH3 128bit hash function
• PseudoRandomGenerator.reset returns self-reference
• added isEmpty method to HyperLogLog and UltraLogLog
• added DistinctCountUtil.deduplicateTokens utility function for the deduplication of hash tokens

Full Changelog: v0.19.0...v0.20.0

v0.19.0

New Features:

• introduced HashStream32::resetAndHashToInt, HashStream64::resetAndHashToLong, and HashStream128:resetAndHashTo128Bits methods for more convenient recycling of hash stream objects
• added HashStream::putUnorderedIterable that takes a HashStream64

Improvements:

• performance optimization of PackedArrayHandler::numEqualComponents for bit size equal to 1

Incompatible Changes:

• removed class XXH3_64 from the public interface, please use corresponding factory method Hashing.xxh3_64 instead
• removed class ConsistentJumpBackBucketHasher from the public interface, please use corresponding factory method ConsistentHashing.jumpBackHash instead

Full Changelog: v0.18.0...v0.19.0

v0.18.0

New Features:

• XXH3: a 64-bit hash function
• added copy method to HashStream #238, thanks to @deripas

Improvements:

• improved HashStream methods for native arrays (putShorts, putInts, ...)

Full Changelog: v0.17.0...v0.18.0

v0.17.0

New Features:

• FarmHashUo: a 64-bit hash function
• JumpBackHash: a very fast consistent hash algorithm

Improvements:

• improved performance of HyperLogLog and UltraLogLog merge operations if sketches have same precision
• better performance of hash functions relying on unsigned 128-bit multiplications on Java 21 by using java.lang.Math.unsignedMultiplyHigh

Full Changelog: v0.16.0...v0.17.0

v0.16.0

Improvements:

• optimized performance of distinct count estimators

Full Changelog: v0.15.0...v0.16.0