Updated changelog [skip ci]

.github/workflows/build.yml

@@ -73,7 +73,6 @@ jobs:
           postgres-version: 14
       - run: |
           call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat" && ^
-          cd %TEMP% && ^
           nmake /NOLOGO /F Makefile.win && ^
           nmake /NOLOGO /F Makefile.win install && ^
           nmake /NOLOGO /F Makefile.win installcheck && ^

CHANGELOG.md

@@ -1,11 +1,11 @@
-## 0.6.2 (2024-03-18)
+## 0.6.2 (unreleased)
 
 - Reduced lock contention with parallel HNSW index builds
 
 ## 0.6.1 (2024-03-04)
 
 - Fixed error with `ANALYZE` and vectors with different dimensions
-- Fixed segmentation fault with `shared_preload_libraries`
+- Fixed error with `shared_preload_libraries`
 - Fixed vector subtraction being marked as commutative
 
 ## 0.6.0 (2024-01-29)

META.json

@@ -2,7 +2,7 @@
 	"name": "vector",
 	"abstract": "Open-source vector similarity search for Postgres",
 	"description": "Supports L2 distance, inner product, and cosine distance",
-	"version": "0.6.2",
+	"version": "0.6.1",
 	"maintainer": [
 		"Andrew Kane <andrew@ankane.org>"
 	],
@@ -20,7 +20,7 @@
 		"vector": {
 			"file": "sql/vector.sql",
 			"docfile": "README.md",
-			"version": "0.6.2",
+			"version": "0.6.1",
 			"abstract": "Open-source vector similarity search for Postgres"
 		}
 	},

Makefile

@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.6.2
+EXTVERSION = 0.6.1
 
 MODULE_big = vector
 DATA = $(wildcard sql/*--*.sql)

Makefile.win

@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.6.2
+EXTVERSION = 0.6.1
 
 OBJS = src\hnsw.obj src\hnswbuild.obj src\hnswinsert.obj src\hnswscan.obj src\hnswutils.obj src\hnswvacuum.obj src\ivfbuild.obj src\ivfflat.obj src\ivfinsert.obj src\ivfkmeans.obj src\ivfscan.obj src\ivfutils.obj src\ivfvacuum.obj src\vector.obj
 HEADERS = src\vector.h

README.md

@@ -20,13 +20,13 @@ Compile and install the extension (supports Postgres 12+)
 
 ```sh
 cd /tmp
-git clone --branch v0.6.2 https://github.com/pgvector/pgvector.git
+git clone --branch v0.6.1 https://github.com/pgvector/pgvector.git
 cd pgvector
 make
 make install # may need sudo
 ```
 
-See the [installation notes](#installation-notes---linux-and-mac) if you run into issues
+See the [installation notes](#installation-notes) if you run into issues
 
 You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#pgxn), [APT](#apt), [Yum](#yum), [pkg](#pkg), or [conda-forge](#conda-forge), and it comes preinstalled with [Postgres.app](#postgresapp) and many [hosted providers](#hosted-postgres). There are also instructions for [GitHub Actions](https://github.com/pgvector/setup-pgvector).
 
@@ -44,15 +44,12 @@ Then use `nmake` to build:
 
 ```cmd
 set "PGROOT=C:\Program Files\PostgreSQL\16"
-cd %TEMP%
-git clone --branch v0.6.2 https://github.com/pgvector/pgvector.git
+git clone --branch v0.6.1 https://github.com/pgvector/pgvector.git
 cd pgvector
 nmake /F Makefile.win
 nmake /F Makefile.win install
 ```
 
-See the [installation notes](#installation-notes---windows) if you run into issues
-
 You can also install it with [Docker](#docker) or [conda-forge](#conda-forge).
 
 ## Getting Started
@@ -413,51 +410,13 @@ You can use [Reciprocal Rank Fusion](https://github.com/pgvector/pgvector-python
 
 ## Performance
 
-### Tuning
-
-Use a tool like [PgTune](https://pgtune.leopard.in.ua/) to set initial values for Postgres server parameters. For instance, `shared_buffers` should typically be 25% of the server’s memory. You can find the config file with:
-
-```sql
-SHOW config_file;
-```
-
-And check individual settings with:
-
-```sql
-SHOW shared_buffers;
-```
-
-Be sure to restart Postgres for changes to take effect.
-
-### Loading
-
-Use `COPY` for bulk loading data ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/bulk_loading.py)).
-
-```sql
-COPY items (embedding) FROM STDIN WITH (FORMAT BINARY);
-```
-
-Add any indexes *after* loading the initial data for best performance.
-
-### Indexing
-
-See index build time for [HNSW](#index-build-time) and [IVFFlat](#index-build-time-1).
-
-In production environments, create indexes concurrently to avoid blocking writes.
-
-```sql
-CREATE INDEX CONCURRENTLY ...
-```
-
-### Querying
-
 Use `EXPLAIN ANALYZE` to debug performance.
 
 ```sql
 EXPLAIN ANALYZE SELECT * FROM items ORDER BY embedding <-> '[3,1,2]' LIMIT 5;
 ```
 
-#### Exact Search
+### Exact Search
 
 To speed up queries without an index, increase `max_parallel_workers_per_gather`.
 
@@ -471,7 +430,7 @@ If vectors are normalized to length 1 (like [OpenAI embeddings](https://platform
 SELECT * FROM items ORDER BY embedding <#> '[3,1,2]' LIMIT 5;
 ```
 
-#### Approximate Search
+### Approximate Search
 
 To speed up queries with an IVFFlat index, increase the number of inverted lists (at the expense of recall).
 
@@ -479,7 +438,7 @@ To speed up queries with an IVFFlat index, increase the number of inverted lists
 CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 1000);
 ```
 
-### Vacuuming
+## Vacuuming
 
 Vacuuming can take a while for HNSW indexes. Speed it up by reindexing first.
 
@@ -488,41 +447,6 @@ REINDEX INDEX CONCURRENTLY index_name;
 VACUUM table_name;
 ```
 
-## Monitoring
-
-Monitor performance with [pg_stat_statements](https://www.postgresql.org/docs/current/pgstatstatements.html) (be sure to add it to `shared_preload_libraries`).
-
-```sql
-CREATE EXTENSION pg_stat_statements;
-```
-
-Get the most time-consuming queries with:
-
-```sql
-SELECT query, calls, ROUND((total_plan_time + total_exec_time) / calls) AS avg_time_ms,
-    ROUND((total_plan_time + total_exec_time) / 60000) AS total_time_min
-    FROM pg_stat_statements ORDER BY total_plan_time + total_exec_time DESC LIMIT 20;
-```
-
-Note: Replace `total_plan_time + total_exec_time` with `total_time` for Postgres < 13
-
-Monitor recall by comparing results from approximate search with exact search.
-
-```sql
-BEGIN;
-SET LOCAL enable_indexscan = off; -- use exact search
-SELECT ...
-COMMIT;
-```
-
-## Scaling
-
-Scale pgvector the same way you scale Postgres.
-
-Scale vertically by increasing memory, CPU, and storage on a single instance. Use existing tools to [tune parameters](#tuning) and [monitor performance](#monitoring).
-
-Scale horizontally with [replicas](https://www.postgresql.org/docs/current/hot-standby.html), or use [Citus](https://github.com/citusdata/citus) or another approach for sharding ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/citus.py)).
-
 ## Languages
 
 Use pgvector from any language with a Postgres client. You can even generate and store vectors in one language and query them in another.
@@ -616,18 +540,6 @@ and query with:
 SELECT * FROM items ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;
 ```
 
-#### Are binary vectors supported?
-
-You can store binary vectors and perform exact nearest neighbor search by Hamming distance in Postgres without an extension ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/hash_image_search.py)).
-
-```tsql
-CREATE TABLE items (id bigserial PRIMARY KEY, embedding bit(3));
-INSERT INTO items (embedding) VALUES (B'000'), (B'111');
-SELECT * FROM items ORDER BY bit_count(embedding # B'101') LIMIT 5;
-```
-
-Indexing is not currently supported.
-
 #### Do indexes need to fit into memory?
 
 No, but like other index types, you’ll likely see better performance if they do. You can get the size of an index with:
@@ -640,17 +552,7 @@ SELECT pg_size_pretty(pg_relation_size('index_name'));
 
 #### Why isn’t a query using an index?
 
-The query needs to have an `ORDER BY` and `LIMIT`, and the `ORDER BY` must be the result of a distance operator, not an expression.
-
-```sql
--- index
-ORDER BY embedding <=> '[3,1,2]' LIMIT 5;
-
--- no index
-ORDER BY 1 - (embedding <=> '[3,1,2]') DESC LIMIT 5;
-```
-
-You can encourage the planner to use an index for a query with:
+The cost estimation in pgvector < 0.4.3 does not always work well with the planner. You can encourage the planner to use an index for a query with:
 
 ```sql
 BEGIN;
@@ -683,8 +585,6 @@ ALTER TABLE items ALTER COLUMN embedding SET STORAGE PLAIN;
 
 Results are limited by the size of the dynamic candidate list (`hnsw.ef_search`). There may be even less results due to dead tuples or filtering conditions in the query. We recommend setting `hnsw.ef_search` to at least twice the `LIMIT` of the query. If you need more than 500 results, use an IVFFlat index instead.
 
-Also, note that `NULL` vectors are not indexed (as well as zero vectors for cosine distance).
-
 #### Why are there less results for a query after adding an IVFFlat index?
 
 The index was likely created with too little data for the number of lists. Drop the index until the table has more data.
@@ -695,13 +595,11 @@ DROP INDEX index_name;
 
 Results can also be limited by the number of probes (`ivfflat.probes`).
 
-Also, note that `NULL` vectors are not indexed (as well as zero vectors for cosine distance).
-
 ## Reference
 
 ### Vector Type
 
-Each vector takes `4 * dimensions + 8` bytes of storage. Each element is a single-precision floating-point number (like the `real` type in Postgres), and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Vectors can have up to 16,000 dimensions.
+Each vector takes `4 * dimensions + 8` bytes of storage. Each element is a single precision floating-point number (like the `real` type in Postgres), and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Vectors can have up to 16,000 dimensions.
 
 ### Vector Operators
 
@@ -725,14 +623,14 @@ l1_distance(vector, vector) → double precision | taxicab distance | 0.5.0
 vector_dims(vector) → integer | number of dimensions |
 vector_norm(vector) → double precision | Euclidean norm |
 
-### Vector Aggregate Functions
+### Aggregate Functions
 
 Function | Description | Added
 --- | --- | ---
 avg(vector) → vector | average |
 sum(vector) → vector | sum | 0.5.0
 
-## Installation Notes - Linux and Mac
+## Installation Notes
 
 ### Postgres Location
 
@@ -774,24 +672,12 @@ If compilation fails and the output includes `warning: no such sysroot directory
 
 ### Portability
 
-By default, pgvector compiles with `-march=native` on some platforms for best performance. However, this can lead to `Illegal instruction` errors if trying to run the compiled extension on a different machine.
-
 To compile for portability, use:
 
 ```sh
 make OPTFLAGS=""
 ```
 
-## Installation Notes - Windows
-
-### Missing Header
-
-If compilation fails with `Cannot open include file: 'postgres.h': No such file or directory`, make sure `PGROOT` is correct.
-
-### Permissions
-
-If installation fails with `Access is denied`, re-run the installation instructions as an administrator.
-
 ## Additional Installation Methods
 
 ### Docker
@@ -807,7 +693,7 @@ This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (r
 You can also build the image manually:
 
 ```sh
-git clone --branch v0.6.2 https://github.com/pgvector/pgvector.git
+git clone --branch v0.6.1 https://github.com/pgvector/pgvector.git
 cd pgvector
 docker build --build-arg PG_MAJOR=16 -t myuser/pgvector .
 ```
@@ -976,12 +862,6 @@ make installcheck REGRESS=functions                            # regression test
 make prove_installcheck PROVE_TESTS=test/t/001_ivfflat_wal.pl  # TAP test
 ```
 
-To enable assertions:
-
-```sh
-make clean && PG_CFLAGS="-DUSE_ASSERT_CHECKING" make && make install
-```
-
 To enable benchmarking:
 
 ```sh
@@ -994,6 +874,12 @@ To show memory usage:
 make clean && PG_CFLAGS="-DHNSW_MEMORY -DIVFFLAT_MEMORY" make && make install
 ```
 
+To enable assertions:
+
+```sh
+make clean && PG_CFLAGS="-DUSE_ASSERT_CHECKING" make && make install
+```
+
 To get k-means metrics:
 
 ```sh

sql/vector--0.6.1--0.6.2.sql

@@ -1,2 +0,0 @@
--- complain if script is sourced in psql, rather than via CREATE EXTENSION
-\echo Use "ALTER EXTENSION vector UPDATE TO '0.6.2'" to load this file. \quit

sql/vector.sql

@@ -1,7 +1,7 @@
 -- complain if script is sourced in psql, rather than via CREATE EXTENSION
 \echo Use "CREATE EXTENSION vector" to load this file. \quit
 
--- vector type
+-- type
 
 CREATE TYPE vector;
 
@@ -29,7 +29,7 @@ CREATE TYPE vector (
 	STORAGE   = external
 );
 
--- vector functions
+-- functions
 
 CREATE FUNCTION l2_distance(vector, vector) RETURNS float8
 	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
@@ -58,7 +58,7 @@ CREATE FUNCTION vector_sub(vector, vector) RETURNS vector
 CREATE FUNCTION vector_mul(vector, vector) RETURNS vector
 	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
 
--- vector private functions
+-- private functions
 
 CREATE FUNCTION vector_lt(vector, vector) RETURNS bool
 	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
@@ -99,7 +99,7 @@ CREATE FUNCTION vector_avg(double precision[]) RETURNS vector
 CREATE FUNCTION vector_combine(double precision[], double precision[]) RETURNS double precision[]
 	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
 
--- vector aggregates
+-- aggregates
 
 CREATE AGGREGATE avg(vector) (
 	SFUNC = vector_accum,
@@ -117,7 +117,7 @@ CREATE AGGREGATE sum(vector) (
 	PARALLEL = SAFE
 );
 
--- vector cast functions
+-- cast functions
 
 CREATE FUNCTION vector(vector, integer, boolean) RETURNS vector
 	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
@@ -137,7 +137,7 @@ CREATE FUNCTION array_to_vector(numeric[], integer, boolean) RETURNS vector
 CREATE FUNCTION vector_to_float4(vector, integer, boolean) RETURNS real[]
 	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
 
--- vector casts
+-- casts
 
 CREATE CAST (vector AS vector)
 	WITH FUNCTION vector(vector, integer, boolean) AS IMPLICIT;
@@ -157,7 +157,7 @@ CREATE CAST (double precision[] AS vector)
 CREATE CAST (numeric[] AS vector)
 	WITH FUNCTION array_to_vector(numeric[], integer, boolean) AS ASSIGNMENT;
 
--- vector operators
+-- operators
 
 CREATE OPERATOR <-> (
 	LEFTARG = vector, RIGHTARG = vector, PROCEDURE = l2_distance,
@@ -240,7 +240,7 @@ CREATE ACCESS METHOD hnsw TYPE INDEX HANDLER hnswhandler;
 
 COMMENT ON ACCESS METHOD hnsw IS 'hnsw index access method';
 
--- vector opclasses
+-- opclasses
 
 CREATE OPERATOR CLASS vector_ops
 	DEFAULT FOR TYPE vector USING btree AS

src/hnsw.h

@@ -55,11 +55,6 @@
 #define HNSW_UPDATE_ENTRY_GREATER 1
 #define HNSW_UPDATE_ENTRY_ALWAYS 2
 
-typedef enum HnswType
-{
-	HNSW_TYPE_VECTOR
-}			HnswType;
-
 /* Build phases */
 /* PROGRESS_CREATEIDX_SUBPHASE_INITIALIZE is 1 */
 #define PROGRESS_HNSW_PHASE_LOAD		2
@@ -134,7 +129,7 @@ HnswPtrDeclare(HnswNeighborArray, HnswNeighborArrayRelptr, HnswNeighborArrayPtr)
 HnswPtrDeclare(HnswNeighborArrayPtr, HnswNeighborsRelptr, HnswNeighborsPtr);
 HnswPtrDeclare(char, DatumRelptr, DatumPtr);
 
-struct HnswElementData
+typedef struct HnswElementData
 {
 	HnswElementPtr next;
 	ItemPointerData heaptids[HNSW_HEAPTIDS];
@@ -149,7 +144,7 @@ struct HnswElementData
 	BlockNumber neighborPage;
 	DatumPtr	value;
 	LWLock		lock;
-};
+}			HnswElementData;
 
 typedef HnswElementData * HnswElement;
 
@@ -160,12 +155,12 @@ typedef struct HnswCandidate
 	bool		closer;
 }			HnswCandidate;
 
-struct HnswNeighborArray
+typedef struct HnswNeighborArray
 {
 	int			length;
 	bool		closerSet;
 	HnswCandidate items[FLEXIBLE_ARRAY_MEMBER];
-};
+}			HnswNeighborArray;
 
 typedef struct HnswPairingHeapNode
 {
@@ -247,7 +242,6 @@ typedef struct HnswBuildState
 	Relation	index;
 	IndexInfo  *indexInfo;
 	ForkNumber	forkNum;
-	HnswType	type;
 
 	/* Settings */
 	int			dimensions;
@@ -268,6 +262,7 @@ typedef struct HnswBuildState
 	HnswGraph  *graph;
 	double		ml;
 	int			maxLevel;
+	Vector	   *normvec;
 
 	/* Memory */
 	MemoryContext graphCtx;
@@ -372,8 +367,7 @@ typedef struct HnswVacuumState
 int			HnswGetM(Relation index);
 int			HnswGetEfConstruction(Relation index);
 FmgrInfo   *HnswOptionalProcInfo(Relation index, uint16 procnum);
-HnswType	HnswGetType(Relation index);
-bool		HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, HnswType type);
+bool		HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
 Buffer		HnswNewBuffer(Relation index, ForkNumber forkNum);
 void		HnswInitPage(Buffer buf, Page page);
 void		HnswInit(void);

src/hnswbuild.c

@@ -436,7 +436,7 @@ InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
 	int			m = buildstate->m;
 	char	   *base = buildstate->hnswarea;
 
-	/* Wait if another process needs exclusive lock on entry lock */
+	/* Wait if another process needs exclusive lock */
 	LWLockAcquire(entryWaitLock, LW_EXCLUSIVE);
 	LWLockRelease(entryWaitLock);
 
@@ -450,7 +450,7 @@ InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
 		/* Release shared lock */
 		LWLockRelease(entryLock);
 
-		/* Tell other processes to wait and get exclusive lock */
+		/* Get exclusive lock */
 		LWLockAcquire(entryWaitLock, LW_EXCLUSIVE);
 		LWLockAcquire(entryLock, LW_EXCLUSIVE);
 		LWLockRelease(entryWaitLock);
@@ -489,7 +489,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
 	/* Normalize if needed */
 	if (buildstate->normprocinfo != NULL)
 	{
-		if (!HnswNormValue(buildstate->normprocinfo, buildstate->collation, &value, buildstate->type))
+		if (!HnswNormValue(buildstate->normprocinfo, buildstate->collation, &value, buildstate->normvec))
 			return false;
 	}
 
@@ -671,13 +671,10 @@ HnswSharedMemoryAlloc(Size size, void *state)
 static void
 InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, IndexInfo *indexInfo, ForkNumber forkNum)
 {
-	int			maxDimensions = HNSW_MAX_DIM;
-
 	buildstate->heap = heap;
 	buildstate->index = index;
 	buildstate->indexInfo = indexInfo;
 	buildstate->forkNum = forkNum;
-	buildstate->type = HnswGetType(index);
 
 	buildstate->m = HnswGetM(index);
 	buildstate->efConstruction = HnswGetEfConstruction(index);
@@ -687,8 +684,8 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
 	if (buildstate->dimensions < 0)
 		elog(ERROR, "column does not have dimensions");
 
-	if (buildstate->dimensions > maxDimensions)
-		elog(ERROR, "column cannot have more than %d dimensions for hnsw index", maxDimensions);
+	if (buildstate->dimensions > HNSW_MAX_DIM)
+		elog(ERROR, "column cannot have more than %d dimensions for hnsw index", HNSW_MAX_DIM);
 
 	if (buildstate->efConstruction < 2 * buildstate->m)
 		elog(ERROR, "ef_construction must be greater than or equal to 2 * m");
@@ -706,6 +703,9 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
 	buildstate->ml = HnswGetMl(buildstate->m);
 	buildstate->maxLevel = HnswGetMaxLevel(buildstate->m);
 
+	/* Reuse for each tuple */
+	buildstate->normvec = InitVector(buildstate->dimensions);
+
 	buildstate->graphCtx = GenerationContextCreate(CurrentMemoryContext,
 												   "Hnsw build graph context",
 #if PG_VERSION_NUM >= 150000
@@ -729,6 +729,7 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
 static void
 FreeBuildState(HnswBuildState * buildstate)
 {
+	pfree(buildstate->normvec);
 	MemoryContextDelete(buildstate->graphCtx);
 	MemoryContextDelete(buildstate->tmpCtx);
 }

src/hnswinsert.c

@@ -622,7 +622,7 @@ HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_ti
 	normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
 	if (normprocinfo != NULL)
 	{
-		if (!HnswNormValue(normprocinfo, collation, &value, HnswGetType(index)))
+		if (!HnswNormValue(normprocinfo, collation, &value, NULL))
 			return;
 	}
 

src/hnswscan.c

@@ -40,6 +40,29 @@ GetScanItems(IndexScanDesc scan, Datum q)
 	return HnswSearchLayer(base, q, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
 }
 
+/*
+ * Get dimensions from metapage
+ */
+static int
+GetDimensions(Relation index)
+{
+	Buffer		buf;
+	Page		page;
+	HnswMetaPage metap;
+	int			dimensions;
+
+	buf = ReadBuffer(index, HNSW_METAPAGE_BLKNO);
+	LockBuffer(buf, BUFFER_LOCK_SHARE);
+	page = BufferGetPage(buf);
+	metap = HnswPageGetMeta(page);
+
+	dimensions = metap->dimensions;
+
+	UnlockReleaseBuffer(buf);
+
+	return dimensions;
+}
+
 /*
  * Get scan value
  */
@@ -50,7 +73,7 @@ GetScanValue(IndexScanDesc scan)
 	Datum		value;
 
 	if (scan->orderByData->sk_flags & SK_ISNULL)
-		value = PointerGetDatum(NULL);
+		value = PointerGetDatum(InitVector(GetDimensions(scan->indexRelation)));
 	else
 	{
 		value = scan->orderByData->sk_argument;
@@ -61,7 +84,7 @@ GetScanValue(IndexScanDesc scan)
 
 		/* Fine if normalization fails */
 		if (so->normprocinfo != NULL)
-			HnswNormValue(so->normprocinfo, so->collation, &value, HnswGetType(scan->indexRelation));
+			HnswNormValue(so->normprocinfo, so->collation, &value, NULL);
 	}
 
 	return value;

src/hnswutils.c

@@ -149,15 +149,6 @@ HnswOptionalProcInfo(Relation index, uint16 procnum)
 	return index_getprocinfo(index, 1, procnum);
 }
 
-/*
- * Get vector type
- */
-HnswType
-HnswGetType(Relation index)
-{
-	return HNSW_TYPE_VECTOR;
-}
-
 /*
  * Divide by the norm
  *
@@ -167,25 +158,21 @@ HnswGetType(Relation index)
  * if it's different than the original value
  */
 bool
-HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, HnswType type)
+HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result)
 {
 	double		norm = DatumGetFloat8(FunctionCall1Coll(procinfo, collation, *value));
 
 	if (norm > 0)
 	{
-		/* TODO Remove vector-specific code */
-		if (type == HNSW_TYPE_VECTOR)
-		{
-			Vector	   *v = DatumGetVector(*value);
-			Vector	   *result = InitVector(v->dim);
+		Vector	   *v = DatumGetVector(*value);
 
-			for (int i = 0; i < v->dim; i++)
-				result->x[i] = v->x[i] / norm;
+		if (result == NULL)
+			result = InitVector(v->dim);
 
-			*value = PointerGetDatum(result);
-		}
-		else
-			elog(ERROR, "Unsupported type");
+		for (int i = 0; i < v->dim; i++)
+			result->x[i] = v->x[i] / norm;
+
+		*value = PointerGetDatum(result);
 
 		return true;
 	}
@@ -588,12 +575,7 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
 
 	/* Calculate distance */
 	if (distance != NULL)
-	{
-		if (DatumGetPointer(*q) == NULL)
-			*distance = 0;
-		else
-			*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
-	}
+		*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
 
 	UnlockReleaseBuffer(buf);
 }

src/ivfbuild.c

@@ -57,7 +57,7 @@ AddSample(Datum *values, IvfflatBuildState * buildstate)
 	 */
 	if (buildstate->kmeansnormprocinfo != NULL)
 	{
-		if (!IvfflatNormValue(buildstate->kmeansnormprocinfo, buildstate->collation, &value))
+		if (!IvfflatNormValue(buildstate->kmeansnormprocinfo, buildstate->collation, &value, buildstate->normvec))
 			return;
 	}
 
@@ -105,7 +105,7 @@ SampleCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
 	oldCtx = MemoryContextSwitchTo(buildstate->tmpCtx);
 
 	/* Add sample */
-	AddSample(values, buildstate);
+	AddSample(values, state);
 
 	/* Reset memory context */
 	MemoryContextSwitchTo(oldCtx);
@@ -153,7 +153,7 @@ AddTupleToSort(Relation index, ItemPointer tid, Datum *values, IvfflatBuildState
 	/* Normalize if needed */
 	if (buildstate->normprocinfo != NULL)
 	{
-		if (!IvfflatNormValue(buildstate->normprocinfo, buildstate->collation, &value))
+		if (!IvfflatNormValue(buildstate->normprocinfo, buildstate->collation, &value, buildstate->normvec))
 			return;
 	}
 
@@ -356,6 +356,9 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
 	buildstate->centers = VectorArrayInit(buildstate->lists, buildstate->dimensions);
 	buildstate->listInfo = palloc(sizeof(ListInfo) * buildstate->lists);
 
+	/* Reuse for each tuple */
+	buildstate->normvec = InitVector(buildstate->dimensions);
+
 	buildstate->tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
 											   "Ivfflat build temporary context",
 											   ALLOCSET_DEFAULT_SIZES);
@@ -377,6 +380,7 @@ FreeBuildState(IvfflatBuildState * buildstate)
 {
 	VectorArrayFree(buildstate->centers);
 	pfree(buildstate->listInfo);
+	pfree(buildstate->normvec);
 
 #ifdef IVFFLAT_KMEANS_DEBUG
 	pfree(buildstate->listSums);

src/ivfflat.h

@@ -172,6 +172,7 @@ typedef struct IvfflatBuildState
 	VectorArray samples;
 	VectorArray centers;
 	ListInfo   *listInfo;
+	Vector	   *normvec;
 
 #ifdef IVFFLAT_KMEANS_DEBUG
 	double		inertia;
@@ -266,7 +267,7 @@ void		VectorArrayFree(VectorArray arr);
 void		PrintVectorArray(char *msg, VectorArray arr);
 void		IvfflatKmeans(Relation index, VectorArray samples, VectorArray centers);
 FmgrInfo   *IvfflatOptionalProcInfo(Relation index, uint16 procnum);
-bool		IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value);
+bool		IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
 int			IvfflatGetLists(Relation index);
 void		IvfflatGetMetaPageInfo(Relation index, int *lists, int *dimensions);
 void		IvfflatUpdateList(Relation index, ListInfo listInfo, BlockNumber insertPage, BlockNumber originalInsertPage, BlockNumber startPage, ForkNumber forkNum);

src/ivfinsert.c

@@ -85,7 +85,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
 	normprocinfo = IvfflatOptionalProcInfo(index, IVFFLAT_NORM_PROC);
 	if (normprocinfo != NULL)
 	{
-		if (!IvfflatNormValue(normprocinfo, index->rd_indcollation[0], &value))
+		if (!IvfflatNormValue(normprocinfo, index->rd_indcollation[0], &value, NULL))
 			return;
 	}
 

src/ivfscan.c

@@ -293,7 +293,7 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
 
 			/* Fine if normalization fails */
 			if (so->normprocinfo != NULL)
-				IvfflatNormValue(so->normprocinfo, so->collation, &value);
+				IvfflatNormValue(so->normprocinfo, so->collation, &value, NULL);
 		}
 
 		IvfflatBench("GetScanLists", GetScanLists(scan, value));

src/ivfutils.c

@@ -75,14 +75,16 @@ IvfflatOptionalProcInfo(Relation index, uint16 procnum)
  * if it's different than the original value
  */
 bool
-IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value)
+IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result)
 {
 	double		norm = DatumGetFloat8(FunctionCall1Coll(procinfo, collation, *value));
 
 	if (norm > 0)
 	{
 		Vector	   *v = DatumGetVector(*value);
-		Vector	   *result = InitVector(v->dim);
+
+		if (result == NULL)
+			result = InitVector(v->dim);
 
 		for (int i = 0; i < v->dim; i++)
 			result->x[i] = v->x[i] / norm;

src/vector.c

@@ -197,8 +197,6 @@ vector_in(PG_FUNCTION_ARGS)
 
 	while (pt != NULL && *stringEnd != ']')
 	{
-		float		val;
-
 		if (dim == VECTOR_MAX_DIM)
 			ereport(ERROR,
 					(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
@@ -214,23 +212,15 @@ vector_in(PG_FUNCTION_ARGS)
 					 errmsg("invalid input syntax for type vector: \"%s\"", lit)));
 
 		/* Use strtof like float4in to avoid a double-rounding problem */
-		errno = 0;
-		val = strtof(pt, &stringEnd);
+		x[dim] = strtof(pt, &stringEnd);
+		CheckElement(x[dim]);
+		dim++;
 
 		if (stringEnd == pt)
 			ereport(ERROR,
 					(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
 					 errmsg("invalid input syntax for type vector: \"%s\"", lit)));
 
-		/* Check for range error like float4in */
-		if (errno == ERANGE && (val == 0 || isinf(val)))
-			ereport(ERROR,
-					(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
-					 errmsg("\"%s\" is out of range for type vector", pt)));
-
-		CheckElement(val);
-		x[dim++] = val;
-
 		while (vector_isspace(*stringEnd))
 			stringEnd++;
 

test/expected/hnsw_l2.out

@@ -12,11 +12,14 @@ SELECT * FROM t ORDER BY val <-> '[3,3,3]';
  [0,0,0]
 (4 rows)
 
-SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector)) t2;
- count 
--------
-     4
-(1 row)
+SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector);
+   val   
+---------
+ [0,0,0]
+ [1,1,1]
+ [1,2,3]
+ [1,2,4]
+(4 rows)
 
 SELECT COUNT(*) FROM t;
  count 

test/expected/input.out

@@ -63,21 +63,9 @@ SELECT '[1.5e-38,-1.5e-38]'::vector;
 (1 row)
 
 SELECT '[4e38,1]'::vector;
-ERROR:  "4e38" is out of range for type vector
+ERROR:  infinite value not allowed in vector
 LINE 1: SELECT '[4e38,1]'::vector;
                ^
-SELECT '[-4e38,1]'::vector;
-ERROR:  "-4e38" is out of range for type vector
-LINE 1: SELECT '[-4e38,1]'::vector;
-               ^
-SELECT '[1e-46,1]'::vector;
-ERROR:  "1e-46" is out of range for type vector
-LINE 1: SELECT '[1e-46,1]'::vector;
-               ^
-SELECT '[-1e-46,1]'::vector;
-ERROR:  "-1e-46" is out of range for type vector
-LINE 1: SELECT '[-1e-46,1]'::vector;
-               ^
 SELECT '[1,2,3'::vector;
 ERROR:  malformed vector literal: "[1,2,3"
 LINE 1: SELECT '[1,2,3'::vector;
@@ -128,30 +116,8 @@ SELECT '[1, ,3]'::vector;
 ERROR:  invalid input syntax for type vector: "[1, ,3]"
 LINE 1: SELECT '[1, ,3]'::vector;
                ^
-SELECT '[1,2,3]'::vector(3);
- vector  
----------
- [1,2,3]
-(1 row)
-
 SELECT '[1,2,3]'::vector(2);
 ERROR:  expected 2 dimensions, not 3
-SELECT '[1,2,3]'::vector(3, 2);
-ERROR:  invalid type modifier
-LINE 1: SELECT '[1,2,3]'::vector(3, 2);
-                          ^
-SELECT '[1,2,3]'::vector('a');
-ERROR:  invalid input syntax for type integer: "a"
-LINE 1: SELECT '[1,2,3]'::vector('a');
-                          ^
-SELECT '[1,2,3]'::vector(0);
-ERROR:  dimensions for type vector must be at least 1
-LINE 1: SELECT '[1,2,3]'::vector(0);
-                          ^
-SELECT '[1,2,3]'::vector(16001);
-ERROR:  dimensions for type vector cannot exceed 16000
-LINE 1: SELECT '[1,2,3]'::vector(16001);
-                          ^
 SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::vector[]);
  unnest  
 ---------

test/sql/hnsw_l2.sql

@@ -7,7 +7,7 @@ CREATE INDEX ON t USING hnsw (val vector_l2_ops);
 INSERT INTO t (val) VALUES ('[1,2,4]');
 
 SELECT * FROM t ORDER BY val <-> '[3,3,3]';
-SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector)) t2;
+SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector);
 SELECT COUNT(*) FROM t;
 
 TRUNCATE t;

test/sql/input.sql

@@ -11,9 +11,6 @@ SELECT '[1.5e38,-1.5e38]'::vector;
 SELECT '[1.5e+38,-1.5e+38]'::vector;
 SELECT '[1.5e-38,-1.5e-38]'::vector;
 SELECT '[4e38,1]'::vector;
-SELECT '[-4e38,1]'::vector;
-SELECT '[1e-46,1]'::vector;
-SELECT '[-1e-46,1]'::vector;
 SELECT '[1,2,3'::vector;
 SELECT '[1,2,3]9'::vector;
 SELECT '1,2,3'::vector;
@@ -25,13 +22,7 @@ SELECT '[1,]'::vector;
 SELECT '[1a]'::vector;
 SELECT '[1,,3]'::vector;
 SELECT '[1, ,3]'::vector;
-
-SELECT '[1,2,3]'::vector(3);
 SELECT '[1,2,3]'::vector(2);
-SELECT '[1,2,3]'::vector(3, 2);
-SELECT '[1,2,3]'::vector('a');
-SELECT '[1,2,3]'::vector(0);
-SELECT '[1,2,3]'::vector(16001);
 
 SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::vector[]);
 SELECT '{"[1,2,3]"}'::vector(2)[];

test/t/016_hnsw_vacuum_recall.pl

@@ -86,7 +86,7 @@ foreach (@queries)
 	push(@expected, $res);
 }
 
-test_recall(0.19, $limit, "before vacuum");
+test_recall(0.20, $limit, "before vacuum");
 test_recall(0.95, 100, "before vacuum");
 
 # TODO Test concurrent inserts with vacuum

vector.control

@@ -1,4 +1,4 @@
 comment = 'vector data type and ivfflat and hnsw access methods'
-default_version = '0.6.2'
+default_version = '0.6.1'
 module_pathname = '$libdir/vector'
 relocatable = true