v1.2.1060

• Add support for concrete values for store, indexing and search
• Improve server error handling and reporting
• Improve cardinality queries in ECL parsing/expansion
• Improve wildcard attribute search

This version does make a change to the backend file-based database. When this version is used against an older version of the database, a warning will be logged as that database will not support provided concrete value results. Any new databases created by this version of hermes will include concrete value support, but older versions of Hermes will ignore those indices. As such, this does not need to be regarded as a breaking change.

Full Changelog: v1.2.1040...v1.2.1060

v1.2.1040

• Improvements to cardinality constraints during ECL processing

Full Changelog: v1.2.1032...v1.2.1040

v1.2.1032

• Optimise handling wildcard value for attribute in an ECL refinement; the prior implementation had no special
  handling for this, and so expanded the value to all concepts, impacting performance.
• Fix handling of cardinality in expressions with a minimum of '0' (e.g. [0..1] or [0..0]).

Full Changelog: v1.2.1026...v1.2.1032

v1.2.1026

• Backwards-compatible changes to core Clojure API with shortened function names (removal of redundant get- prefix)
• Miscellaneous optimisations/speed-ups, including in building queries for ECL expansion
• Internal (private API) changes
• Improved, and more complete tests
• More deterministic derivation of 'replaced by' relationship in the graph API
• Remove unused dependencies from uberjar, speeding up startup

Full Changelog: v1.1.1000...v1.2.1026

v1.1.1000

• Update to version 1.1 series given drop in support for Lucene 8.x series and therefore Java 1.8
• Upgrade dependencies (e.g. netty-buffer 4.1.90)
• Add ability to stream search results asynchronously
• Miscellaneous performance improvements
• Miscellaneous code style improvements

Full Changelog: v1.0.960...v1.1.1000

v1.0.960

• Drop support for Lucene 8.x series and therefore Java 1.8 (latter out of support from 31 March 2022, and Lucene 9.x requires at least Java 11).
• Improved Java API with thin wrapper API for use by Java and other JVM languages when used as a library
• Add ability to stream all concepts to top level API
• Use new Lucene 9.5 'storedFields' in favour of deprecated document field access
• Add new benchmark test suite
• Optimisations
• Upgrade internal dependencies (Lucene 9.5, lmdbjava 0.8.3)

v1.0.938

• Add support to give multiple commands in one invocation via command-line interface
• Improve speed of import through tweaking memory map synchronisation
• Improve speed of indexing through more efficient calculation of transitive closure tables for relationships

v1.0.921

Full Changelog: v1.0.914...v1.0.921

v1.0.910

This release has mainly internal changes compared to prior release. The only user-facing change is exposing status* as part of core API.