Advanced and operational features

Return to the feature comparison overview for status definitions and comparison scope.

Vector and hybrid search

Feature	LeanCorpus	Lucene.NET 4.8	Java Lucene	Notes
Float-vector fields and k-nearest-neighbour queries	✔	❌	✔	HNSW-backed approximate search.
Filtered approximate kNN	✔	❌	✔	Pre-filter and post-filter modes.
HNSW build configuration	✔	❌	◐	Per-index graph parameters and deterministic seed support.
Vector normalisation	✔	❌	◐	Optional index-time L2 normalisation.
Int8 and binary vector quantisation	✔	❌	✔	Scalar and binary quantisation across indexing and search.
Four-bit and product quantisation	❌	❌	✔	Not currently available.
SIMD vector operations	✔	❌	◐	Runtime-vectorised .NET implementation using Vector<T>.
Reciprocal-rank fusion	✔	❌	◐	Native query form for lexical and vector result fusion.
Byte-vector fields and queries	❌	❌	✔	Float vectors are supported instead.
Similarity-threshold vector query	❌	❌	✔	No dedicated threshold query.

Facets, aggregations, and suggestions

Feature	LeanCorpus	Lucene.NET 4.8	Java Lucene	Notes
Term facets	✔	✔	✔	Sorted or sorted-set DocValues with bounded exact bucket collection.
Numeric min, max, sum, average, and count	✔	◐	◐	First-class numeric aggregation requests.
Fixed-bucket histograms	✔	❌	❌	LeanCorpus-specific aggregation.
Range facets	✔	✔	✔	Explicit numeric and UTC date ranges with inclusive/exclusive bounds.
Taxonomy and hierarchical facets	◐	✔	✔	Explicit FacetPath prefixes use existing StringField postings and sorted-set DocValues; no taxonomy index.
Drill-down and drill-sideways	✔	✔	✔	AND across dimensions, OR within a dimension, with one base traversal for sideways counts.
Approximate cardinality aggregation	✔	❌	❌	Bounded HLL-style sparse/dense numeric cardinality aggregation.
Percentile aggregation	✔	❌	❌	t-digest doubles and HDR-style logarithmic non-negative Int64 percentile aggregations.
Spell checking	✔	✔	✔	DidYouMeanSuggester backed by a spell index.
Prefix suggestions	✔	✔	✔	FST completion ranked by global document frequency.
Analysing suggestions	✔	✔	✔	Applies the selected analyser to completion input.
Free-text suggestions	✔	✔	✔	Uses analysed phrase context to suggest the next term.
Context-filtered suggestions	◐	✔	✔	LeanCorpus filters analysing suggestions with a query rather than exposing Lucene's context-suggester API.
Fuzzy suggestions	◐	✔	✔	LeanCorpus provides edit-distance spelling suggestions rather than Lucene's weighted fuzzy completion API.

Geo and spatial search

Feature	LeanCorpus	Lucene.NET 4.8	Java Lucene	Notes
Geo point indexing	◐	◐	✔	GeoPointField uses a BKD-backed representation with a different API.
Bounding-box queries	✔	✔	✔	Latitude and longitude box matching.
Distance queries	✔	✔	✔	Radius matching around a point.
Geo encoding utilities	✔	✔	✔	Coordinate encoding helpers.
Polygon, line, and geo-shape queries	❌	◐	✔	Lucene.NET offers older Spatial4n strategies.
Cartesian point and shape search	❌	❌	✔	No XY point or shape API.

Search controls and extensions

Feature	LeanCorpus	Lucene.NET 4.8	Java Lucene	Notes
Per-query timeout	✔	✔	✔	Cooperative checks between segments.
Cancellation tokens	✔	❌	◐	Native .NET cancellation semantics.
Per-query memory budget	✔	❌	❌	Caps intermediate result memory.
Partial-result signalling	✔	❌	❌	TopDocs.IsPartial records budget or timeout truncation.
Asynchronous result streaming	✔	❌	❌	IAsyncEnumerable<ScoreDoc> search.
Segment-by-segment streaming	✔	❌	❌	Supports pipeline consumption before all segments complete.
Named ranking profiles and query rules	✔	❌	❌	Immutable relevance configuration, bounded pipelines, filters, boosts, and pins.
Snapshot-bound cursor sessions	✔	◐	◐	Opaque cursors remain bound to a retained committed searcher view.
Document classification	❌	✔	✔	No classifier module.
Percolation and stored-query monitoring	❌	✔	✔	No MemoryIndex or monitor-query equivalent.

Diagnostics, tooling, and deployment

Feature	LeanCorpus	Lucene.NET 4.8	Java Lucene	Notes
OpenTelemetry traces	✔	❌	❌	ActivitySource spans across search and index operations.
.NET metrics	✔	❌	❌	Counters and histograms for index maintenance.
Slow-query records and search analytics	✔	❌	❌	Bounded in-process diagnostic buffers.
Typed index and segment reports	✔	◐	◐	Higher-level reports over file and segment metadata.
Index check and inspection CLI	✔	◐	◐	Structured output and JSON support.
Backup, compatibility, and migration CLI	✔	❌	❌	Commands over LeanCorpus-native operational APIs.
Desktop index browser	❌	✔	✔	Lucene provides Luke.
Native AOT compatibility	✔	❌	❌	Trim-safe core without runtime code generation.
Source-generated serialisation metadata	✔	❌	❌	Reflection-free System.Text.Json metadata.