Added support for uuid

These places called PageIndexTupleOverwrite(), with the new tuple
pointing directly to the original page. The PageIndexTupleOverwrite()
call is unnecessary in these cases, as we have already modified the
tuple on the page directly. Moreover, PageIndexTupleOverWrite() will
call memcpy with same src and dst arguments, which is undefined
behavior.

It's OK to modify the pages on disk directly, and we don't need
critical sections, because we either use the generic xlog functions
which create a temporary copy of the page, or we are building a new
index so if we crash the whole index is invisible and will be dropped
anyway.

.github/workflows/build.yml

@@ -57,9 +57,9 @@ jobs:
       - run: |
           brew install cpanm
           cpanm --notest IPC::Run
-          wget -q https://github.com/postgres/postgres/archive/refs/tags/REL_14_5.tar.gz
-          tar xf REL_14_5.tar.gz
-      - run: make prove_installcheck PROVE_FLAGS="-I ./postgres-REL_14_5/src/test/perl" PERL5LIB="/Users/runner/perl5/lib/perl5"
+          wget -q https://github.com/postgres/postgres/archive/refs/tags/REL_14_10.tar.gz
+          tar xf REL_14_10.tar.gz
+      - run: make prove_installcheck PROVE_FLAGS="-I ./postgres-REL_14_10/src/test/perl" PERL5LIB="/Users/runner/perl5/lib/perl5"
       - run: make clean && /usr/local/opt/llvm@15/bin/scan-build --status-bugs make PG_CFLAGS="-DUSE_ASSERT_CHECKING"
   windows:
     runs-on: windows-latest

CHANGELOG.md

@@ -1,16 +1,25 @@
-## 0.6.0 (unreleased)
+## 0.7.0 (unreleased)
 
-If upgrading with Postgres < 13, see [this note](https://github.com/pgvector/pgvector#060).
+- Added support for inline filtering with HNSW
 
+## 0.6.1 (unreleased)
+
+- Fixed error with `ANALYZE` and vectors with different dimensions
+
+## 0.6.0 (2024-01-29)
+
+If upgrading with Postgres 12 or Docker, see [these notes](https://github.com/pgvector/pgvector#060).
+
+- Added support for parallel index builds for HNSW
+- Added validation for GUC parameters
 - Changed storage for vector from `extended` to `external`
 - Improved performance of HNSW
-- Added support for parallel index builds for HNSW
-- Added `hamming_distance` function
-- Added validation for GUC parameters
 - Reduced memory usage for HNSW index builds
 - Reduced WAL generation for HNSW index builds
 - Fixed error with logical replication
-- Fixed `invalid memory alloc request size` error with HNSW index build
+- Fixed `invalid memory alloc request size` error with HNSW index builds
+- Moved Docker image to `pgvector` org
+- Added Docker tags for each supported version of Postgres
 - Dropped support for Postgres 11
 
 ## 0.5.1 (2023-10-10)
@@ -59,7 +68,7 @@ If upgrading with Postgres < 13, see [this note](https://github.com/pgvector/pgv
 
 ## 0.4.0 (2023-01-11)
 
-If upgrading with Postgres < 13, see [this note](https://github.com/pgvector/pgvector#040).
+If upgrading with Postgres < 13, see [this note](https://github.com/pgvector/pgvector/blob/v0.4.0/README.md#040).
 
 - Changed text representation for vector elements to match `real`
 - Changed storage for vector from `plain` to `extended`
@@ -76,7 +85,7 @@ If upgrading with Postgres < 13, see [this note](https://github.com/pgvector/pgv
 
 ## 0.3.1 (2022-11-02)
 
-If upgrading from 0.2.7 or 0.3.0, [recreate](https://github.com/pgvector/pgvector#031) all `ivfflat` indexes after upgrading to ensure all data is indexed.
+If upgrading from 0.2.7 or 0.3.0, [recreate](https://github.com/pgvector/pgvector/blob/v0.3.1/README.md#031) all `ivfflat` indexes after upgrading to ensure all data is indexed.
 
 - Fixed issue with inserts silently corrupting `ivfflat` indexes (introduced in 0.2.7)
 - Fixed segmentation fault with index creation when lists > 6500

Dockerfile

@@ -1,4 +1,4 @@
-ARG PG_MAJOR=15
+ARG PG_MAJOR=16
 FROM postgres:$PG_MAJOR
 ARG PG_MAJOR
 

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": "0.5.1",
+	"version": "0.6.0",
 	"maintainer": [
 		"Andrew Kane <andrew@ankane.org>"
 	],
@@ -20,7 +20,7 @@
 		"vector": {
 			"file": "sql/vector.sql",
 			"docfile": "README.md",
-			"version": "0.5.1",
+			"version": "0.6.0",
 			"abstract": "Open-source vector similarity search for Postgres"
 		}
 	},

Makefile

@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.5.1
+EXTVERSION = 0.6.0
 
 MODULE_big = vector
 DATA = $(wildcard sql/*--*.sql)
@@ -65,13 +65,17 @@ dist:
 	mkdir -p dist
 	git archive --format zip --prefix=$(EXTENSION)-$(EXTVERSION)/ --output dist/$(EXTENSION)-$(EXTVERSION).zip master
 
+# for Docker
+PG_MAJOR ?= 16
+
 .PHONY: docker
 
 docker:
-	docker build --pull --no-cache --platform linux/amd64 -t ankane/pgvector:latest .
+	docker build --pull --no-cache --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR) .
+	docker build --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) .
 
 .PHONY: docker-release
 
 docker-release:
-	docker buildx build --push --pull --no-cache --platform linux/amd64,linux/arm64 -t ankane/pgvector:latest .
-	docker buildx build --push --platform linux/amd64,linux/arm64 -t ankane/pgvector:v$(EXTVERSION) .
+	docker buildx build --push --pull --no-cache --platform linux/amd64,linux/arm64 --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR) .
+	docker buildx build --push --platform linux/amd64,linux/arm64 --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) .

Makefile.win

@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.5.1
+EXTVERSION = 0.6.0
 
 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

@@ -14,11 +14,13 @@ Plus [ACID](https://en.wikipedia.org/wiki/ACID) compliance, point-in-time recove
 
 ## Installation
 
-Compile and install the extension (supports Postgres 11+)
+### Linux and Mac
+
+Compile and install the extension (supports Postgres 12+)
 
 ```sh
 cd /tmp
-git clone --branch v0.5.1 https://github.com/pgvector/pgvector.git
+git clone --branch v0.6.0 https://github.com/pgvector/pgvector.git
 cd pgvector
 make
 make install # may need sudo
@@ -26,7 +28,29 @@ make install # may need sudo
 
 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), 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).
+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).
+
+### Windows
+
+Ensure [C++ support in Visual Studio](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#download-and-install-the-tools) is installed, and run:
+
+```cmd
+call "C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvars64.bat"
+```
+
+Note: The exact path will vary depending on your Visual Studio version and edition
+
+Then use `nmake` to build:
+
+```cmd
+set "PGROOT=C:\Program Files\PostgreSQL\16"
+git clone --branch v0.6.0 https://github.com/pgvector/pgvector.git
+cd pgvector
+nmake /F Makefile.win
+nmake /F Makefile.win install
+```
+
+You can also install it with [Docker](#docker) or [conda-forge](#conda-forge).
 
 ## Getting Started
 
@@ -240,6 +264,14 @@ HINT:  Increase maintenance_work_mem to speed up builds.
 
 Note: Do not set `maintenance_work_mem` so high that it exhausts the memory on the server
 
+Starting with 0.6.0, you can also speed up index creation by increasing the number of parallel workers (2 by default)
+
+```sql
+SET max_parallel_maintenance_workers = 7; -- plus leader
+```
+
+For a large number of workers, you may also need to increase `max_parallel_workers` (8 by default)
+
 ### Indexing Progress
 
 Check [indexing progress](https://www.postgresql.org/docs/current/progress-reporting.html#CREATE-INDEX-PROGRESS-REPORTING) with Postgres 12+
@@ -345,6 +377,12 @@ Create an index on one [or more](https://www.postgresql.org/docs/current/indexes
 CREATE INDEX ON items (category_id);
 ```
 
+Or a composite HNSW index for approximate search (added in 0.7.0)
+
+```sql
+CREATE INDEX ON items USING hnsw (embedding vector_l2_ops, category_id);
+```
+
 Or a [partial index](https://www.postgresql.org/docs/current/indexes-partial.html) on the vector column for approximate search
 
 ```sql
@@ -398,6 +436,15 @@ 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 can take a while for HNSW indexes. Speed it up by reindexing first.
+
+```sql
+REINDEX INDEX CONCURRENTLY index_name;
+VACUUM table_name;
+```
+
 ## 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.
@@ -615,44 +662,24 @@ Note: Replace `16` with your Postgres server version
 
 If compilation fails and the output includes `warning: no such sysroot directory` on Mac, reinstall Xcode Command Line Tools.
 
-### Windows
-
-Support for Windows is currently experimental. Ensure [C++ support in Visual Studio](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#download-and-install-the-tools) is installed, and run:
-
-```cmd
-call "C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvars64.bat"
-```
-
-Note: The exact path will vary depending on your Visual Studio version and edition
-
-Then use `nmake` to build:
-
-```cmd
-set "PGROOT=C:\Program Files\PostgreSQL\16"
-git clone --branch v0.5.1 https://github.com/pgvector/pgvector.git
-cd pgvector
-nmake /F Makefile.win
-nmake /F Makefile.win install
-```
-
 ## Additional Installation Methods
 
 ### Docker
 
-Get the [Docker image](https://hub.docker.com/r/ankane/pgvector) with:
+Get the [Docker image](https://hub.docker.com/r/pgvector/pgvector) with:
 
 ```sh
-docker pull ankane/pgvector
+docker pull pgvector/pgvector:pg16
 ```
 
-This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (run it the same way).
+This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (replace `16` with your Postgres server version, and run it the same way).
 
 You can also build the image manually:
 
 ```sh
-git clone --branch v0.5.1 https://github.com/pgvector/pgvector.git
+git clone --branch v0.6.0 https://github.com/pgvector/pgvector.git
 cd pgvector
-docker build --build-arg PG_MAJOR=15 -t myuser/pgvector .
+docker build --build-arg PG_MAJOR=16 -t myuser/pgvector .
 ```
 
 ### Homebrew
@@ -695,6 +722,21 @@ sudo dnf install pgvector_16
 
 Note: Replace `16` with your Postgres server version
 
+### pkg
+
+Install the FreeBSD package with:
+
+```sh
+pkg install postgresql15-pg_vector
+```
+
+or the port with:
+
+```sh
+cd /usr/ports/databases/pg_vector
+make install
+```
+
 ### conda-forge
 
 With Conda Postgres, install from [conda-forge](https://anaconda.org/conda-forge/pgvector) with:
@@ -729,9 +771,11 @@ SELECT extversion FROM pg_extension WHERE extname = 'vector';
 
 ## Upgrade Notes
 
-### 0.6.0 [unreleased]
+### 0.6.0
 
-If upgrading with Postgres < 13, remove this line from `sql/vector--0.5.1--0.6.0.sql`:
+#### Postgres 12
+
+If upgrading with Postgres 12, remove this line from `sql/vector--0.5.1--0.6.0.sql`:
 
 ```sql
 ALTER TYPE vector SET (STORAGE = external);
@@ -739,28 +783,20 @@ ALTER TYPE vector SET (STORAGE = external);
 
 Then run `make install` and `ALTER EXTENSION vector UPDATE;`.
 
-### 0.4.0
+#### Docker
 
-If upgrading with Postgres < 13, remove this line from `sql/vector--0.3.2--0.4.0.sql`:
+The Docker image is now published in the `pgvector` org, and there are tags for each supported version of Postgres (rather than a `latest` tag).
 
-```sql
-ALTER TYPE vector SET (STORAGE = extended);
+```sh
+docker pull pgvector/pgvector:pg16
+# or
+docker pull pgvector/pgvector:0.6.0-pg16
 ```
 
-Then run `make install` and `ALTER EXTENSION vector UPDATE;`.
+Also, if you’ve increased `maintenance_work_mem`, make sure `--shm-size` is at least that size to avoid an error with parallel HNSW index builds.
 
-### 0.3.1
-
-If upgrading from 0.2.7 or 0.3.0, recreate all `ivfflat` indexes after upgrading to ensure all data is indexed.
-
-```sql
--- Postgres 12+
-REINDEX INDEX CONCURRENTLY index_name;
-
--- Postgres < 12
-CREATE INDEX CONCURRENTLY temp_name ON table USING ivfflat (column opclass);
-DROP INDEX CONCURRENTLY index_name;
-ALTER INDEX temp_name RENAME TO index_name;
+```sh
+docker run --shm-size=1g ...
 ```
 
 ## Thanks
@@ -806,8 +842,8 @@ make prove_installcheck  # TAP tests
 To run single tests:
 
 ```sh
-make installcheck REGRESS=functions                    # regression test
-make prove_installcheck PROVE_TESTS=test/t/001_wal.pl  # TAP test
+make installcheck REGRESS=functions                            # regression test
+make prove_installcheck PROVE_TESTS=test/t/001_ivfflat_wal.pl  # TAP test
 ```
 
 To enable benchmarking:

sql/vector--0.5.1--0.6.0.sql

@@ -3,6 +3,3 @@
 
 -- remove this single line for Postgres < 13
 ALTER TYPE vector SET (STORAGE = external);
-
-CREATE FUNCTION hamming_distance(bytea, bytea) RETURNS float8
-	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;

sql/vector--0.6.1--0.7.0.sql

@@ -0,0 +1,34 @@
+-- complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "ALTER EXTENSION vector UPDATE TO '0.7.0'" to load this file. \quit
+
+CREATE OPERATOR CLASS vector_bigint_ops
+	DEFAULT FOR TYPE bigint USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;
+
+CREATE OPERATOR CLASS vector_integer_ops
+	DEFAULT FOR TYPE integer USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;
+
+CREATE OPERATOR CLASS vector_text_ops
+	DEFAULT FOR TYPE text USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;
+
+CREATE OPERATOR CLASS vector_uuid_ops
+	DEFAULT FOR TYPE uuid USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;

sql/vector.sql

@@ -291,7 +291,36 @@ CREATE OPERATOR CLASS vector_cosine_ops
 	FUNCTION 1 vector_negative_inner_product(vector, vector),
 	FUNCTION 2 vector_norm(vector);
 
--- bytea functions
+-- hnsw attributes
 
-CREATE FUNCTION hamming_distance(bytea, bytea) RETURNS float8
-	AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
+CREATE OPERATOR CLASS vector_bigint_ops
+	DEFAULT FOR TYPE bigint USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;
+
+CREATE OPERATOR CLASS vector_integer_ops
+	DEFAULT FOR TYPE integer USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;
+
+CREATE OPERATOR CLASS vector_text_ops
+	DEFAULT FOR TYPE text USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;
+
+CREATE OPERATOR CLASS vector_uuid_ops
+	DEFAULT FOR TYPE uuid USING hnsw AS
+	OPERATOR 2 < ,
+	OPERATOR 3 <= ,
+	OPERATOR 4 = ,
+	OPERATOR 5 >= ,
+	OPERATOR 6 > ;

src/hnsw.c

@@ -200,7 +200,7 @@ hnswhandler(PG_FUNCTION_ARGS)
 	amroutine->amcanorderbyop = true;
 	amroutine->amcanbackward = false;	/* can change direction mid-scan */
 	amroutine->amcanunique = false;
-	amroutine->amcanmulticol = false;
+	amroutine->amcanmulticol = true;
 	amroutine->amoptionalkey = true;
 	amroutine->amsearcharray = false;
 	amroutine->amsearchnulls = false;

src/hnsw.h

@@ -60,10 +60,13 @@
 #define PROGRESS_HNSW_PHASE_LOAD		2
 
 #define HNSW_MAX_SIZE (BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(HnswPageOpaqueData)) - sizeof(ItemIdData))
+#define HNSW_TUPLE_ALLOC_SIZE BLCKSZ
 
 #define HNSW_ELEMENT_TUPLE_SIZE(size)	MAXALIGN(offsetof(HnswElementTupleData, data) + (size))
 #define HNSW_NEIGHBOR_TUPLE_SIZE(level, m)	MAXALIGN(offsetof(HnswNeighborTupleData, indextids) + ((level) + 2) * (m) * sizeof(ItemPointerData))
 
+#define HNSW_NEIGHBOR_ARRAY_SIZE(lm)	(offsetof(HnswNeighborArray, items) + sizeof(HnswCandidate) * (lm))
+
 #define HnswPageGetOpaque(page)	((HnswPageOpaque) PageGetSpecialPointer(page))
 #define HnswPageGetMeta(page)	((HnswMetaPageData *) PageGetContents(page))
 
@@ -125,6 +128,7 @@ HnswPtrDeclare(HnswElementData, HnswElementRelptr, HnswElementPtr);
 HnswPtrDeclare(HnswNeighborArray, HnswNeighborArrayRelptr, HnswNeighborArrayPtr);
 HnswPtrDeclare(HnswNeighborArrayPtr, HnswNeighborsRelptr, HnswNeighborsPtr);
 HnswPtrDeclare(char, DatumRelptr, DatumPtr);
+HnswPtrDeclare(IndexTupleData, IndexTupleRelptr, IndexTuplePtr);
 
 typedef struct HnswElementData
 {
@@ -140,6 +144,7 @@ typedef struct HnswElementData
 	OffsetNumber neighborOffno;
 	BlockNumber neighborPage;
 	DatumPtr	value;
+	IndexTuplePtr itup;
 	LWLock		lock;
 }			HnswElementData;
 
@@ -367,7 +372,7 @@ bool		HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * re
 Buffer		HnswNewBuffer(Relation index, ForkNumber forkNum);
 void		HnswInitPage(Buffer buf, Page page);
 void		HnswInit(void);
-List	   *HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement);
+List	   *HnswSearchLayer(char *base, Datum q, IndexScanDesc scan, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement);
 HnswElement HnswGetEntryPoint(Relation index);
 void		HnswGetMetaPageInfo(Relation index, int *m, HnswElement * entryPoint);
 void	   *HnswAlloc(HnswAllocator * allocator, Size size);
@@ -379,13 +384,16 @@ void		HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint
 void		HnswSetNeighborTuple(char *base, HnswNeighborTuple ntup, HnswElement e, int m);
 void		HnswAddHeapTid(HnswElement element, ItemPointer heaptid);
 void		HnswInitNeighbors(char *base, HnswElement element, int m, HnswAllocator * alloc);
-bool		HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel, bool building);
+bool		HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull, ItemPointer heap_tid, bool building);
 void		HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting, bool building);
-void		HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec);
-void		HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec);
-void		HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element);
+void		HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec, Relation index);
+void		HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec, IndexScanDesc scan, bool *matches);
+void		HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element, bool useIndexTuple);
 void		HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, FmgrInfo *procinfo, Oid collation);
 void		HnswLoadNeighbors(HnswElement element, Relation index, int m);
+TupleDesc	HnswTupleDesc(Relation index);
+IndexTuple	HnswFormIndexTuple(Relation index, TupleDesc tupdesc, Datum value, Datum *values, bool *isnull);
+bool		HnswElementIsDuplicate(char *base, HnswElement a, HnswElement b, Relation index);
 PGDLLEXPORT void HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc);
 
 /* Index access methods */

src/hnswbuild.c

@@ -65,8 +65,6 @@
 #define CALLBACK_ITEM_POINTER HeapTuple hup
 #endif
 
-#define UpdateProgress(index, val) pgstat_progress_update_param(index, val)
-
 #if PG_VERSION_NUM >= 140000
 #include "utils/backend_status.h"
 #include "utils/wait_event.h"
@@ -143,14 +141,13 @@ HnswBuildAppendPage(Relation index, Buffer *buf, Page *page, ForkNumber forkNum)
 }
 
 /*
- * Create element pages
+ * Create graph pages
  */
 static void
-CreateElementPages(HnswBuildState * buildstate)
+CreateGraphPages(HnswBuildState * buildstate)
 {
 	Relation	index = buildstate->index;
 	ForkNumber	forkNum = buildstate->forkNum;
-	Size		etupAllocSize;
 	Size		maxSize;
 	HnswElementTuple etup;
 	HnswNeighborTuple ntup;
@@ -160,14 +157,14 @@ CreateElementPages(HnswBuildState * buildstate)
 	Page		page;
 	HnswElementPtr iter = buildstate->graph->head;
 	char	   *base = buildstate->hnswarea;
+	bool		useIndexTuple = IndexRelationGetNumberOfAttributes(index) > 1;
 
 	/* Calculate sizes */
-	etupAllocSize = BLCKSZ;
 	maxSize = HNSW_MAX_SIZE;
 
 	/* Allocate once */
-	etup = palloc0(etupAllocSize);
-	ntup = palloc0(BLCKSZ);
+	etup = palloc0(HNSW_TUPLE_ALLOC_SIZE);
+	ntup = palloc0(HNSW_TUPLE_ALLOC_SIZE);
 
 	/* Prepare first page */
 	buf = HnswNewBuffer(index, forkNum);
@@ -180,24 +177,25 @@ CreateElementPages(HnswBuildState * buildstate)
 		Size		etupSize;
 		Size		ntupSize;
 		Size		combinedSize;
-		void	   *valuePtr = HnswPtrAccess(base, element->value);
+		Pointer		valuePtr = HnswPtrAccess(base, element->value);
+		IndexTuple	itup = HnswPtrAccess(base, element->itup);
 
 		/* Update iterator */
 		iter = element->next;
 
 		/* Zero memory for each element */
-		MemSet(etup, 0, etupAllocSize);
+		MemSet(etup, 0, HNSW_TUPLE_ALLOC_SIZE);
 
 		/* Calculate sizes */
-		etupSize = HNSW_ELEMENT_TUPLE_SIZE(VARSIZE_ANY(valuePtr));
+		etupSize = HNSW_ELEMENT_TUPLE_SIZE(useIndexTuple ? IndexTupleSize(itup) : VARSIZE_ANY(valuePtr));
 		ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(element->level, buildstate->m);
 		combinedSize = etupSize + ntupSize + sizeof(ItemIdData);
 
 		/* Initial size check */
-		if (etupSize > etupAllocSize)
+		if (etupSize > HNSW_TUPLE_ALLOC_SIZE)
 			elog(ERROR, "index tuple too large");
 
-		HnswSetElementTuple(base, etup, element);
+		HnswSetElementTuple(base, etup, element, useIndexTuple);
 
 		/* Keep element and neighbors on the same page if possible */
 		if (PageGetFreeSpace(page) < etupSize || (combinedSize <= maxSize && PageGetFreeSpace(page) < combinedSize))
@@ -246,10 +244,10 @@ CreateElementPages(HnswBuildState * buildstate)
 }
 
 /*
- * Create neighbor pages
+ * Write neighbor tuples
  */
 static void
-CreateNeighborPages(HnswBuildState * buildstate)
+WriteNeighborTuples(HnswBuildState * buildstate)
 {
 	Relation	index = buildstate->index;
 	ForkNumber	forkNum = buildstate->forkNum;
@@ -259,29 +257,32 @@ CreateNeighborPages(HnswBuildState * buildstate)
 	HnswNeighborTuple ntup;
 
 	/* Allocate once */
-	ntup = palloc0(BLCKSZ);
+	ntup = palloc0(HNSW_TUPLE_ALLOC_SIZE);
 
 	while (!HnswPtrIsNull(base, iter))
 	{
-		HnswElement e = HnswPtrAccess(base, iter);
+		HnswElement element = HnswPtrAccess(base, iter);
 		Buffer		buf;
 		Page		page;
-		Size		ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(e->level, m);
+		Size		ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(element->level, m);
 
 		/* Update iterator */
-		iter = e->next;
+		iter = element->next;
+
+		/* Zero memory for each element */
+		MemSet(ntup, 0, HNSW_TUPLE_ALLOC_SIZE);
 
 		/* Can take a while, so ensure we can interrupt */
 		/* Needs to be called when no buffer locks are held */
 		CHECK_FOR_INTERRUPTS();
 
-		buf = ReadBufferExtended(index, forkNum, e->neighborPage, RBM_NORMAL, NULL);
+		buf = ReadBufferExtended(index, forkNum, element->neighborPage, RBM_NORMAL, NULL);
 		LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
 		page = BufferGetPage(buf);
 
-		HnswSetNeighborTuple(base, ntup, e, m);
+		HnswSetNeighborTuple(base, ntup, element, m);
 
-		if (!PageIndexTupleOverwrite(page, e->neighborOffno, (Item) ntup, ntupSize))
+		if (!PageIndexTupleOverwrite(page, element->neighborOffno, (Item) ntup, ntupSize))
 			elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
 
 		/* Commit */
@@ -303,8 +304,8 @@ FlushPages(HnswBuildState * buildstate)
 #endif
 
 	CreateMetaPage(buildstate);
-	CreateElementPages(buildstate);
-	CreateNeighborPages(buildstate);
+	CreateGraphPages(buildstate);
+	WriteNeighborTuples(buildstate);
 
 	buildstate->graph->flushed = true;
 	MemoryContextReset(buildstate->graphCtx);
@@ -335,19 +336,17 @@ AddDuplicateInMemory(HnswElement element, HnswElement dup)
  * Find duplicate element
  */
 static bool
-FindDuplicateInMemory(char *base, HnswElement element)
+FindDuplicateInMemory(char *base, HnswElement element, Relation index)
 {
 	HnswNeighborArray *neighbors = HnswGetNeighbors(base, element, 0);
-	Datum		value = HnswGetValue(base, element);
 
 	for (int i = 0; i < neighbors->length; i++)
 	{
 		HnswCandidate *neighbor = &neighbors->items[i];
 		HnswElement neighborElement = HnswPtrAccess(base, neighbor->element);
-		Datum		neighborValue = HnswGetValue(base, neighborElement);
 
 		/* Exit early since ordered by distance */
-		if (!datumIsEqual(value, neighborValue, false, -1))
+		if (!HnswElementIsDuplicate(base, element, neighborElement, index))
 			return false;
 
 		/* Check for space */
@@ -407,7 +406,7 @@ UpdateGraphInMemory(FmgrInfo *procinfo, Oid collation, HnswElement element, int
 	char	   *base = buildstate->hnswarea;
 
 	/* Look for duplicate */
-	if (FindDuplicateInMemory(base, element))
+	if (FindDuplicateInMemory(base, element, buildstate->index))
 		return;
 
 	/* Add element */
@@ -472,10 +471,14 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
 	HnswGraph  *graph = buildstate->graph;
 	HnswElement element;
 	HnswAllocator *allocator = &buildstate->allocator;
-	Size		valueSize;
+	IndexTuple	itup;
+	Size		itupSize;
+	IndexTuple	itupPtr;
 	Pointer		valuePtr;
 	LWLock	   *flushLock = &graph->flushLock;
 	char	   *base = buildstate->hnswarea;
+	TupleDesc	tupdesc = HnswTupleDesc(index);
+	bool		unused;
 
 	/* Detoast once for all calls */
 	Datum		value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
@@ -488,7 +491,8 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
 	}
 
 	/* Get datum size */
-	valueSize = VARSIZE_ANY(DatumGetPointer(value));
+	itup = HnswFormIndexTuple(index, tupdesc, value, values, isnull);
+	itupSize = IndexTupleSize(itup);
 
 	/* Ensure graph not flushed when inserting */
 	LWLockAcquire(flushLock, LW_SHARED);
@@ -498,7 +502,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
 	{
 		LWLockRelease(flushLock);
 
-		return HnswInsertTupleOnDisk(index, value, values, isnull, heaptid, buildstate->heap, true);
+		return HnswInsertTupleOnDisk(index, value, values, isnull, heaptid, true);
 	}
 
 	/*
@@ -530,12 +534,12 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
 
 		LWLockRelease(flushLock);
 
-		return HnswInsertTupleOnDisk(index, value, values, isnull, heaptid, buildstate->heap, true);
+		return HnswInsertTupleOnDisk(index, value, values, isnull, heaptid, true);
 	}
 
 	/* Ok, we can proceed to allocate the element */
 	element = HnswInitElement(base, heaptid, buildstate->m, buildstate->ml, buildstate->maxLevel, allocator);
-	valuePtr = HnswAlloc(allocator, valueSize);
+	itupPtr = HnswAlloc(allocator, itupSize);
 
 	/*
 	 * We have now allocated the space needed for the element, so we don't
@@ -544,8 +548,10 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
 	 */
 	LWLockRelease(&graph->allocatorLock);
 
-	/* Copy the datum */
-	memcpy(valuePtr, DatumGetPointer(value), valueSize);
+	/* Copy the index tuple */
+	memcpy(itupPtr, itup, itupSize);
+	HnswPtrStore(base, element->itup, itupPtr);
+	valuePtr = DatumGetPointer(index_getattr(itupPtr, 1, tupdesc, &unused));
 	HnswPtrStore(base, element->value, valuePtr);
 
 	/* Create a lock for the element */
@@ -587,7 +593,7 @@ BuildCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
 	{
 		/* Update progress */
 		SpinLockAcquire(&graph->lock);
-		UpdateProgress(PROGRESS_CREATEIDX_TUPLES_DONE, ++graph->indtuples);
+		pgstat_progress_update_param(PROGRESS_CREATEIDX_TUPLES_DONE, ++graph->indtuples);
 		SpinLockRelease(&graph->lock);
 	}
 
@@ -670,6 +676,19 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
 	buildstate->efConstruction = HnswGetEfConstruction(index);
 	buildstate->dimensions = TupleDescAttr(index->rd_att, 0)->atttypmod;
 
+	/* For now */
+	if (IndexRelationGetNumberOfKeyAttributes(index) > 3)
+		elog(ERROR, "index cannot have more than three columns");
+
+	if (!OidIsValid(index_getprocid(index, 1, HNSW_DISTANCE_PROC)))
+		elog(ERROR, "first column must be a vector");
+
+	for (int i = 1; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
+	{
+		if (OidIsValid(index_getprocid(index, i + 1, HNSW_DISTANCE_PROC)))
+			elog(ERROR, "column %d cannot be a vector", i + 1);
+	}
+
 	/* Require column to have dimensions to be indexed */
 	if (buildstate->dimensions < 0)
 		elog(ERROR, "column does not have dimensions");
@@ -1048,7 +1067,7 @@ BuildGraph(HnswBuildState * buildstate, ForkNumber forkNum)
 {
 	int			parallel_workers = 0;
 
-	UpdateProgress(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_HNSW_PHASE_LOAD);
+	pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_HNSW_PHASE_LOAD);
 
 	/* Calculate parallel workers */
 	if (buildstate->heap != NULL)

src/hnswinsert.c

@@ -136,9 +136,12 @@ AddElementOnDisk(Relation index, HnswElement e, int m, BlockNumber insertPage, B
 	OffsetNumber freeNeighborOffno = InvalidOffsetNumber;
 	BlockNumber newInsertPage = InvalidBlockNumber;
 	char	   *base = NULL;
+	bool		useIndexTuple = IndexRelationGetNumberOfAttributes(index) > 1;
+	Pointer		valuePtr = HnswPtrAccess(base, e->value);
+	IndexTuple	itup = HnswPtrAccess(base, e->itup);
 
 	/* Calculate sizes */
-	etupSize = HNSW_ELEMENT_TUPLE_SIZE(VARSIZE_ANY(HnswPtrAccess(base, e->value)));
+	etupSize = HNSW_ELEMENT_TUPLE_SIZE(useIndexTuple ? IndexTupleSize(itup) : VARSIZE_ANY(valuePtr));
 	ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(e->level, m);
 	combinedSize = etupSize + ntupSize + sizeof(ItemIdData);
 	maxSize = HNSW_MAX_SIZE;
@@ -146,7 +149,7 @@ AddElementOnDisk(Relation index, HnswElement e, int m, BlockNumber insertPage, B
 
 	/* Prepare element tuple */
 	etup = palloc0(etupSize);
-	HnswSetElementTuple(base, etup, e);
+	HnswSetElementTuple(base, etup, e, useIndexTuple);
 
 	/* Prepare neighbor tuple */
 	ntup = palloc0(ntupSize);
@@ -355,9 +358,7 @@ HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, Hns
 			Buffer		buf;
 			Page		page;
 			GenericXLogState *state;
-			ItemId		itemid;
 			HnswNeighborTuple ntup;
-			Size		ntupSize;
 			int			idx = -1;
 			int			startIdx;
 			HnswElement neighborElement = HnswPtrAccess(base, hc->element);
@@ -396,9 +397,7 @@ HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, Hns
 			}
 
 			/* Get tuple */
-			itemid = PageGetItemId(page, offno);
-			ntup = (HnswNeighborTuple) PageGetItem(page, itemid);
-			ntupSize = ItemIdGetLength(itemid);
+			ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, offno));
 
 			/* Calculate index for update */
 			startIdx = (neighborElement->level - lc) * m;
@@ -427,13 +426,9 @@ HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, Hns
 			{
 				ItemPointer indextid = &ntup->indextids[idx];
 
-				/* Update neighbor */
+				/* Update neighbor on the buffer */
 				ItemPointerSet(indextid, e->blkno, e->offno);
 
-				/* Overwrite tuple */
-				if (!PageIndexTupleOverwrite(page, offno, (Item) ntup, ntupSize))
-					elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
-
 				/* Commit */
 				if (building)
 					MarkBufferDirty(buf);
@@ -457,9 +452,7 @@ AddDuplicateOnDisk(Relation index, HnswElement element, HnswElement dup, bool bu
 	Buffer		buf;
 	Page		page;
 	GenericXLogState *state;
-	ItemId		itemid;
 	HnswElementTuple etup;
-	Size		etupSize;
 	int			i;
 
 	/* Read page */
@@ -477,9 +470,7 @@ AddDuplicateOnDisk(Relation index, HnswElement element, HnswElement dup, bool bu
 	}
 
 	/* Find space */
-	itemid = PageGetItemId(page, dup->offno);
-	etup = (HnswElementTuple) PageGetItem(page, itemid);
-	etupSize = ItemIdGetLength(itemid);
+	etup = (HnswElementTuple) PageGetItem(page, PageGetItemId(page, dup->offno));
 	for (i = 0; i < HNSW_HEAPTIDS; i++)
 	{
 		if (!ItemPointerIsValid(&etup->heaptids[i]))
@@ -495,13 +486,9 @@ AddDuplicateOnDisk(Relation index, HnswElement element, HnswElement dup, bool bu
 		return false;
 	}
 
-	/* Add heap TID */
+	/* Add heap TID, modifying the tuple on the page directly */
 	etup->heaptids[i] = element->heaptids[0];
 
-	/* Overwrite tuple */
-	if (!PageIndexTupleOverwrite(page, dup->offno, (Item) etup, etupSize))
-		elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
-
 	/* Commit */
 	if (building)
 		MarkBufferDirty(buf);
@@ -520,16 +507,14 @@ FindDuplicateOnDisk(Relation index, HnswElement element, bool building)
 {
 	char	   *base = NULL;
 	HnswNeighborArray *neighbors = HnswGetNeighbors(base, element, 0);
-	Datum		value = HnswGetValue(base, element);
 
 	for (int i = 0; i < neighbors->length; i++)
 	{
 		HnswCandidate *neighbor = &neighbors->items[i];
 		HnswElement neighborElement = HnswPtrAccess(base, neighbor->element);
-		Datum		neighborValue = HnswGetValue(base, neighborElement);
 
 		/* Exit early since ordered by distance */
-		if (!datumIsEqual(value, neighborValue, false, -1))
+		if (!HnswElementIsDuplicate(base, element, neighborElement, index))
 			return false;
 
 		if (AddDuplicateOnDisk(index, element, neighborElement, building))
@@ -570,7 +555,7 @@ UpdateGraphOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement
  * Insert a tuple into the index
  */
 bool
-HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel, bool building)
+HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull, ItemPointer heap_tid, bool building)
 {
 	HnswElement entryPoint;
 	HnswElement element;
@@ -580,6 +565,10 @@ HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull,
 	Oid			collation = index->rd_indcollation[0];
 	LOCKMODE	lockmode = ShareLock;
 	char	   *base = NULL;
+	TupleDesc	tupdesc = HnswTupleDesc(index);
+	bool		unused;
+	IndexTuple	itup;
+	Pointer		valuePtr;
 
 	/*
 	 * Get a shared lock. This allows vacuum to ensure no in-flight inserts
@@ -593,7 +582,10 @@ HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull,
 
 	/* Create an element */
 	element = HnswInitElement(base, heap_tid, m, HnswGetMl(m), HnswGetMaxLevel(m), NULL);
-	HnswPtrStore(base, element->value, DatumGetPointer(value));
+	itup = HnswFormIndexTuple(index, tupdesc, value, values, isnull);
+	HnswPtrStore(base, element->itup, itup);
+	valuePtr = DatumGetPointer(index_getattr(itup, 1, tupdesc, &unused));
+	HnswPtrStore(base, element->value, valuePtr);
 
 	/* Prevent concurrent inserts when likely updating entry point */
 	if (entryPoint == NULL || element->level > entryPoint->level)
@@ -625,7 +617,7 @@ HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull,
  * Insert a tuple into the index
  */
 static void
-HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel)
+HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid)
 {
 	Datum		value;
 	FmgrInfo   *normprocinfo;
@@ -642,7 +634,7 @@ HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_ti
 			return;
 	}
 
-	HnswInsertTupleOnDisk(index, value, values, isnull, heap_tid, heapRel, false);
+	HnswInsertTupleOnDisk(index, value, values, isnull, heap_tid, false);
 }
 
 /*
@@ -671,7 +663,7 @@ hnswinsert(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid,
 	oldCtx = MemoryContextSwitchTo(insertCtx);
 
 	/* Insert tuple */
-	HnswInsertTuple(index, values, isnull, heap_tid, heap);
+	HnswInsertTuple(index, values, isnull, heap_tid);
 
 	/* Delete memory context */
 	MemoryContextSwitchTo(oldCtx);

src/hnswscan.c

@@ -33,11 +33,12 @@ GetScanItems(IndexScanDesc scan, Datum q)
 
 	for (int lc = entryPoint->level; lc >= 1; lc--)
 	{
-		w = HnswSearchLayer(base, q, ep, 1, lc, index, procinfo, collation, m, false, NULL);
+		w = HnswSearchLayer(base, q, NULL, ep, 1, lc, index, procinfo, collation, m, false, NULL);
 		ep = w;
 	}
 
-	return HnswSearchLayer(base, q, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
+	/* Only pass scan to check matches for layer 0 */
+	return HnswSearchLayer(base, q, scan, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
 }
 
 /*
@@ -202,6 +203,7 @@ hnswgettuple(IndexScanDesc scan, ScanDirection dir)
 		MemoryContextSwitchTo(oldCtx);
 
 		scan->xs_heaptid = *heaptid;
+		scan->xs_recheck = scan->numberOfKeys > 0;
 		scan->xs_recheckorderby = false;
 		return true;
 	}

src/hnswutils.c

@@ -3,13 +3,19 @@
 #include <math.h>
 
 #include "access/generic_xlog.h"
+#include "access/relscan.h"
 #include "hnsw.h"
 #include "lib/pairingheap.h"
 #include "storage/bufmgr.h"
 #include "utils/datum.h"
+#include "utils/memdebug.h"
 #include "utils/rel.h"
 #include "vector.h"
 
+#if PG_VERSION_NUM < 130000
+#define TYPSTORAGE_PLAIN 'p'
+#endif
+
 #if PG_VERSION_NUM >= 130000
 #include "common/hashfn.h"
 #else
@@ -202,6 +208,19 @@ HnswInitPage(Buffer buf, Page page)
 	HnswPageGetOpaque(page)->page_id = HNSW_PAGE_ID;
 }
 
+/*
+ * Allocate a neighbor array
+ */
+static HnswNeighborArray *
+HnswInitNeighborArray(int lm, HnswAllocator * allocator)
+{
+	HnswNeighborArray *a = HnswAlloc(allocator, HNSW_NEIGHBOR_ARRAY_SIZE(lm));
+
+	a->length = 0;
+	a->closerSet = false;
+	return a;
+}
+
 /*
  * Allocate neighbors
  */
@@ -209,22 +228,12 @@ void
 HnswInitNeighbors(char *base, HnswElement element, int m, HnswAllocator * allocator)
 {
 	int			level = element->level;
-
 	HnswNeighborArrayPtr *neighborList = (HnswNeighborArrayPtr *) HnswAlloc(allocator, sizeof(HnswNeighborArrayPtr) * (level + 1));
 
 	HnswPtrStore(base, element->neighbors, neighborList);
 
 	for (int lc = 0; lc <= level; lc++)
-	{
-		HnswNeighborArray *a;
-		int			lm = HnswGetLayerM(m, lc);
-
-		HnswPtrStore(base, neighborList[lc], (HnswNeighborArray *) HnswAlloc(allocator, offsetof(HnswNeighborArray, items) + sizeof(HnswCandidate) * lm));
-
-		a = HnswGetNeighbors(base, element, lc);
-		a->length = 0;
-		a->closerSet = false;
-	}
+		HnswPtrStore(base, neighborList[lc], HnswInitNeighborArray(HnswGetLayerM(m, lc), allocator));
 }
 
 /*
@@ -291,6 +300,82 @@ HnswInitElementFromBlock(BlockNumber blkno, OffsetNumber offno)
 	return element;
 }
 
+/*
+ * Get the tuple descriptor
+ */
+TupleDesc
+HnswTupleDesc(Relation index)
+{
+	TupleDesc	tupdesc = CreateTupleDescCopyConstr(RelationGetDescr(index));
+
+	/* Prevent compression */
+	TupleDescAttr(tupdesc, 0)->attstorage = TYPSTORAGE_PLAIN;
+
+	return tupdesc;
+}
+
+/*
+ * Form an index tuple
+ */
+IndexTuple
+HnswFormIndexTuple(Relation index, TupleDesc tupdesc, Datum value, Datum *values, bool *isnull)
+{
+	Size		size = sizeof(Datum) * IndexRelationGetNumberOfAttributes(index);
+	Datum	   *newValues = palloc(size);
+
+	memcpy(newValues, values, size);
+	newValues[0] = value;
+
+	return index_form_tuple(tupdesc, newValues, isnull);
+}
+
+/*
+ * Check if elements are duplicates
+ */
+bool
+HnswElementIsDuplicate(char *base, HnswElement a, HnswElement b, Relation index)
+{
+	if (IndexRelationGetNumberOfAttributes(index) == 1)
+	{
+		Datum		value = HnswGetValue(base, a);
+		Datum		value2 = HnswGetValue(base, b);
+
+		return datumIsEqual(value, value2, false, -1);
+	}
+	else
+	{
+		TupleDesc	tupdesc = RelationGetDescr(index);
+		IndexTuple	itup = HnswPtrAccess(base, a->itup);
+		IndexTuple	itup2 = HnswPtrAccess(base, b->itup);
+
+		for (int i = 0; i < tupdesc->natts; i++)
+		{
+			Datum		value;
+			Datum		value2;
+			bool		isnull;
+			bool		isnull2;
+
+			value = index_getattr(itup, i + 1, tupdesc, &isnull);
+			value2 = index_getattr(itup2, i + 1, tupdesc, &isnull2);
+
+			if (isnull || isnull2)
+			{
+				if (isnull != isnull2)
+					return false;
+			}
+			else
+			{
+				Form_pg_attribute att = TupleDescAttr(tupdesc, i);
+
+				if (!datumIsEqual(value, value2, att->attbyval, att->attlen))
+					return false;
+			}
+		}
+
+		return true;
+	}
+}
+
 /*
  * Get the metapage info
  */
@@ -312,7 +397,10 @@ HnswGetMetaPageInfo(Relation index, int *m, HnswElement * entryPoint)
 	if (entryPoint != NULL)
 	{
 		if (BlockNumberIsValid(metap->entryBlkno))
+		{
 			*entryPoint = HnswInitElementFromBlock(metap->entryBlkno, metap->entryOffno);
+			(*entryPoint)->level = metap->entryLevel;
+		}
 		else
 			*entryPoint = NULL;
 	}
@@ -397,10 +485,8 @@ HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, Bloc
  * Set element tuple, except for neighbor info
  */
 void
-HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element)
+HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element, bool useIndexTuple)
 {
-	Pointer		valuePtr = HnswPtrAccess(base, element->value);
-
 	etup->type = HNSW_ELEMENT_TUPLE_TYPE;
 	etup->level = element->level;
 	etup->deleted = 0;
@@ -411,7 +497,19 @@ HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element)
 		else
 			ItemPointerSetInvalid(&etup->heaptids[i]);
 	}
-	memcpy(&etup->data, valuePtr, VARSIZE_ANY(valuePtr));
+
+	if (useIndexTuple)
+	{
+		IndexTuple	itup = HnswPtrAccess(base, element->itup);
+
+		memcpy(&etup->data, itup, IndexTupleSize(itup));
+	}
+	else
+	{
+		Pointer		valuePtr = HnswPtrAccess(base, element->value);
+
+		memcpy(&etup->data, valuePtr, VARSIZE_ANY(valuePtr));
+	}
 }
 
 /*
@@ -515,7 +613,7 @@ HnswLoadNeighbors(HnswElement element, Relation index, int m)
  * Load an element from a tuple
  */
 void
-HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec)
+HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec, Relation index)
 {
 	element->level = etup->level;
 	element->deleted = etup->deleted;
@@ -538,17 +636,64 @@ HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHe
 	if (loadVec)
 	{
 		char	   *base = NULL;
-		Datum		value = datumCopy(PointerGetDatum(&etup->data), false, -1);
 
-		HnswPtrStore(base, element->value, DatumGetPointer(value));
+		if (IndexRelationGetNumberOfAttributes(index) > 1)
+		{
+			TupleDesc	tupdesc = RelationGetDescr(index);
+			bool		unused;
+			IndexTuple	itup = CopyIndexTuple((IndexTuple) &etup->data);
+			Datum		value = index_getattr(itup, 1, tupdesc, &unused);
+
+			HnswPtrStore(base, element->itup, itup);
+			HnswPtrStore(base, element->value, DatumGetPointer(value));
+		}
+		else
+		{
+			Datum		value = datumCopy(PointerGetDatum(&etup->data), false, -1);
+
+			HnswPtrStore(base, element->value, DatumGetPointer(value));
+		}
 	}
 }
 
+/*
+ * Check if an element matches
+ */
+static bool
+HnswCheckMatches(Relation index, HnswElementTuple etup, IndexScanDesc scan)
+{
+	if (scan == NULL)
+		return true;
+
+	for (int i = 0; i < scan->numberOfKeys; i++)
+	{
+		IndexTuple	itup = (IndexTuple) &etup->data;
+		TupleDesc	tupdesc = RelationGetDescr(index);
+		ScanKey		key = &scan->keyData[i];
+		bool		isnull;
+		Datum		value = index_getattr(itup, key->sk_attno, tupdesc, &isnull);
+		bool		attnull = key->sk_flags & SK_ISNULL;
+
+		if (isnull || attnull)
+		{
+			if (isnull != attnull)
+				return false;
+		}
+		else
+		{
+			if (!DatumGetBool(FunctionCall2Coll(&key->sk_func, key->sk_collation, value, key->sk_argument)))
+				return false;
+		}
+	}
+
+	return true;
+}
+
 /*
  * Load an element and optionally get its distance from q
  */
 void
-HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec)
+HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec, IndexScanDesc scan, bool *matches)
 {
 	Buffer		buf;
 	Page		page;
@@ -564,11 +709,29 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
 	Assert(HnswIsElementTuple(etup));
 
 	/* Load element */
-	HnswLoadElementFromTuple(element, etup, true, loadVec);
+	HnswLoadElementFromTuple(element, etup, true, loadVec, index);
 
 	/* Calculate distance */
 	if (distance != NULL)
-		*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
+	{
+		Datum		value;
+
+		if (IndexRelationGetNumberOfAttributes(index) > 1)
+		{
+			IndexTuple	itup = (IndexTuple) &etup->data;
+			TupleDesc	tupdesc = RelationGetDescr(index);
+			bool		unused;
+
+			value = index_getattr(itup, 1, tupdesc, &unused);
+		}
+		else
+			value = PointerGetDatum(&etup->data);
+
+		*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, value));
+	}
+
+	if (matches != NULL)
+		*matches = HnswCheckMatches(index, etup, scan);
 
 	UnlockReleaseBuffer(buf);
 }
@@ -597,7 +760,7 @@ HnswEntryCandidate(char *base, HnswElement entryPoint, Datum q, Relation index,
 	if (index == NULL)
 		hc->distance = GetCandidateDistance(base, hc, q, procinfo, collation);
 	else
-		HnswLoadElement(entryPoint, &hc->distance, &q, index, procinfo, collation, loadVec);
+		HnswLoadElement(entryPoint, &hc->distance, &q, index, procinfo, collation, loadVec, NULL, NULL);
 	return hc;
 }
 
@@ -715,12 +878,14 @@ CountElement(char *base, HnswElement skipElement, HnswCandidate * hc)
  * Algorithm 2 from paper
  */
 List *
-HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement)
+HnswSearchLayer(char *base, Datum q, IndexScanDesc scan, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement)
 {
 	List	   *w = NIL;
 	pairingheap *C = pairingheap_allocate(CompareNearestCandidates, NULL);
 	pairingheap *W = pairingheap_allocate(CompareFurthestCandidates, NULL);
 	int			wlen = 0;
+	uint64		additional = 0;
+	uint64		maxAdditional = (scan != NULL && scan->numberOfKeys > 0) ? 100 * ef : 0;
 	visited_hash v;
 	ListCell   *lc2;
 	HnswNeighborArray *neighborhoodData = NULL;
@@ -731,7 +896,7 @@ HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, F
 	/* Create local memory for neighborhood if needed */
 	if (index == NULL)
 	{
-		neighborhoodSize = offsetof(HnswNeighborArray, items) + sizeof(HnswCandidate) * HnswGetLayerM(m, lc);
+		neighborhoodSize = HNSW_NEIGHBOR_ARRAY_SIZE(HnswGetLayerM(m, lc));
 		neighborhoodData = palloc(neighborhoodSize);
 	}
 
@@ -792,14 +957,18 @@ HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, F
 			if (!visited)
 			{
 				float		eDistance;
+				bool		eMatches;
 				HnswElement eElement = HnswPtrAccess(base, e->element);
 
 				f = ((HnswPairingHeapNode *) pairingheap_first(W))->inner;
 
 				if (index == NULL)
+				{
 					eDistance = GetCandidateDistance(base, e, q, procinfo, collation);
+					eMatches = true;
+				}
 				else
-					HnswLoadElement(eElement, &eDistance, &q, index, procinfo, collation, inserting);
+					HnswLoadElement(eElement, &eDistance, &q, index, procinfo, collation, inserting, scan, &eMatches);
 
 				Assert(!eElement->deleted);
 
@@ -818,6 +987,16 @@ HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, F
 					pairingheap_add(C, &(CreatePairingHeapNode(ec)->ph_node));
 					pairingheap_add(W, &(CreatePairingHeapNode(ec)->ph_node));
 
+					/*
+					 * Do not count elements that do not match filter towards
+					 * ef
+					 */
+					if (!eMatches)
+					{
+						if ((++additional) <= maxAdditional)
+							continue;
+					}
+
 					/*
 					 * Do not count elements being deleted towards ef when
 					 * vacuuming. It would be ideal to do this for inserts as
@@ -907,40 +1086,10 @@ CompareCandidateDistancesOffset(const void *a, const void *b)
  * Calculate the distance between elements
  */
 static float
-HnswGetDistance(char *base, HnswElement a, HnswElement b, int lc, FmgrInfo *procinfo, Oid collation)
+HnswGetDistance(char *base, HnswElement a, HnswElement b, FmgrInfo *procinfo, Oid collation)
 {
-	Datum		aValue;
-	Datum		bValue;
-
-	/* Look for cached distance */
-	if (!HnswPtrIsNull(base, a->neighbors))
-	{
-		HnswNeighborArray *neighbors = HnswGetNeighbors(base, a, lc);
-
-		for (int i = 0; i < neighbors->length; i++)
-		{
-			HnswElement element = HnswPtrAccess(base, neighbors->items[i].element);
-
-			if (element == b)
-				return neighbors->items[i].distance;
-		}
-	}
-
-	if (!HnswPtrIsNull(base, b->neighbors))
-	{
-		HnswNeighborArray *neighbors = HnswGetNeighbors(base, b, lc);
-
-		for (int i = 0; i < neighbors->length; i++)
-		{
-			HnswElement element = HnswPtrAccess(base, neighbors->items[i].element);
-
-			if (element == a)
-				return neighbors->items[i].distance;
-		}
-	}
-
-	aValue = HnswGetValue(base, a);
-	bValue = HnswGetValue(base, b);
+	Datum		aValue = HnswGetValue(base, a);
+	Datum		bValue = HnswGetValue(base, b);
 
 	return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, aValue, bValue));
 }
@@ -949,7 +1098,7 @@ HnswGetDistance(char *base, HnswElement a, HnswElement b, int lc, FmgrInfo *proc
  * Check if an element is closer to q than any element from R
  */
 static bool
-CheckElementCloser(char *base, HnswCandidate * e, List *r, int lc, FmgrInfo *procinfo, Oid collation)
+CheckElementCloser(char *base, HnswCandidate * e, List *r, FmgrInfo *procinfo, Oid collation)
 {
 	HnswElement eElement = HnswPtrAccess(base, e->element);
 	ListCell   *lc2;
@@ -958,7 +1107,7 @@ CheckElementCloser(char *base, HnswCandidate * e, List *r, int lc, FmgrInfo *pro
 	{
 		HnswCandidate *ri = lfirst(lc2);
 		HnswElement riElement = HnswPtrAccess(base, ri->element);
-		float		distance = HnswGetDistance(base, eElement, riElement, lc, procinfo, collation);
+		float		distance = HnswGetDistance(base, eElement, riElement, procinfo, collation);
 
 		if (distance <= e->distance)
 			return false;
@@ -1004,16 +1153,20 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
 
 		/* Use previous state of r and wd to skip work when possible */
 		if (mustCalculate)
-			e->closer = CheckElementCloser(base, e, r, lc, procinfo, collation);
+			e->closer = CheckElementCloser(base, e, r, procinfo, collation);
 		else if (list_length(added) > 0)
 		{
+			/* Keep Valgrind happy for in-memory, parallel builds */
+			if (base != NULL)
+				VALGRIND_MAKE_MEM_DEFINED(&e->closer, 1);
+
 			/*
 			 * If the current candidate was closer, we only need to compare it
 			 * with the other candidates that we have added.
 			 */
 			if (e->closer)
 			{
-				e->closer = CheckElementCloser(base, e, added, lc, procinfo, collation);
+				e->closer = CheckElementCloser(base, e, added, procinfo, collation);
 
 				if (!e->closer)
 					removedAny = true;
@@ -1026,7 +1179,7 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
 				 */
 				if (removedAny)
 				{
-					e->closer = CheckElementCloser(base, e, r, lc, procinfo, collation);
+					e->closer = CheckElementCloser(base, e, r, procinfo, collation);
 					if (e->closer)
 						added = lappend(added, e);
 				}
@@ -1034,11 +1187,15 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
 		}
 		else if (e == newCandidate)
 		{
-			e->closer = CheckElementCloser(base, e, r, lc, procinfo, collation);
+			e->closer = CheckElementCloser(base, e, r, procinfo, collation);
 			if (e->closer)
 				added = lappend(added, e);
 		}
 
+		/* Keep Valgrind happy for in-memory, parallel builds */
+		if (base != NULL)
+			VALGRIND_MAKE_MEM_DEFINED(&e->closer, 1);
+
 		if (e->closer)
 			r = lappend(r, e);
 		else
@@ -1114,7 +1271,7 @@ HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm
 				HnswElement hc3Element = HnswPtrAccess(base, hc3->element);
 
 				if (HnswPtrIsNull(base, hc3Element->value))
-					HnswLoadElement(hc3Element, &hc3->distance, &q, index, procinfo, collation, true);
+					HnswLoadElement(hc3Element, &hc3->distance, &q, index, procinfo, collation, true, NULL, NULL);
 				else
 					hc3->distance = GetCandidateDistance(base, hc3, q, procinfo, collation);
 
@@ -1236,7 +1393,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
 	/* 1st phase: greedy search to insert level */
 	for (int lc = entryLevel; lc >= level + 1; lc--)
 	{
-		w = HnswSearchLayer(base, q, ep, 1, lc, index, procinfo, collation, m, true, skipElement);
+		w = HnswSearchLayer(base, q, NULL, ep, 1, lc, index, procinfo, collation, m, true, skipElement);
 		ep = w;
 	}
 
@@ -1254,7 +1411,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
 		List	   *neighbors;
 		List	   *lw;
 
-		w = HnswSearchLayer(base, q, ep, efConstruction, lc, index, procinfo, collation, m, true, skipElement);
+		w = HnswSearchLayer(base, q, NULL, ep, efConstruction, lc, index, procinfo, collation, m, true, skipElement);
 
 		/* Elements being deleted or skipped can help with search */
 		/* but should be removed before selecting neighbors */

src/hnswvacuum.c

@@ -60,8 +60,7 @@ RemoveHeapTids(HnswVacuumState * vacuumstate)
 		/* Iterate over nodes */
 		for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
 		{
-			ItemId		itemid = PageGetItemId(page, offno);
-			HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, itemid);
+			HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, PageGetItemId(page, offno));
 			int			idx = 0;
 			bool		itemUpdated = false;
 
@@ -92,15 +91,10 @@ RemoveHeapTids(HnswVacuumState * vacuumstate)
 
 				if (itemUpdated)
 				{
-					Size		etupSize = ItemIdGetLength(itemid);
-
 					/* Mark rest as invalid */
 					for (int i = idx; i < HNSW_HEAPTIDS; i++)
 						ItemPointerSetInvalid(&etup->heaptids[i]);
 
-					if (!PageIndexTupleOverwrite(page, offno, (Item) etup, etupSize))
-						elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
-
 					updated = true;
 				}
 			}
@@ -213,6 +207,9 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
 	/* Find neighbors for element, skipping itself */
 	HnswFindElementNeighbors(base, element, entryPoint, index, procinfo, collation, m, efConstruction, true);
 
+	/* Zero memory for each element */
+	MemSet(ntup, 0, HNSW_TUPLE_ALLOC_SIZE);
+
 	/* Update neighbor tuple */
 	/* Do this before getting page to minimize locking */
 	HnswSetNeighborTuple(base, ntup, element, m);
@@ -259,7 +256,7 @@ RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
 		LockPage(index, HNSW_UPDATE_LOCK, ShareLock);
 
 		/* Load element */
-		HnswLoadElement(highestPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true);
+		HnswLoadElement(highestPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true, NULL, NULL);
 
 		/* Repair if needed */
 		if (NeedsUpdated(vacuumstate, highestPoint))
@@ -297,7 +294,7 @@ RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
 			 * is outdated, this can remove connections at higher levels in
 			 * the graph until they are repaired, but this should be fine.
 			 */
-			HnswLoadElement(entryPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true);
+			HnswLoadElement(entryPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true, NULL, NULL);
 
 			if (NeedsUpdated(vacuumstate, entryPoint))
 			{
@@ -373,7 +370,7 @@ RepairGraph(HnswVacuumState * vacuumstate)
 
 			/* Create an element */
 			element = HnswInitElementFromBlock(blkno, offno);
-			HnswLoadElementFromTuple(element, etup, false, true);
+			HnswLoadElementFromTuple(element, etup, false, true, index);
 
 			elements = lappend(elements, element);
 		}
@@ -443,6 +440,7 @@ MarkDeleted(HnswVacuumState * vacuumstate)
 	BlockNumber insertPage = InvalidBlockNumber;
 	Relation	index = vacuumstate->index;
 	BufferAccessStrategy bas = vacuumstate->bas;
+	bool		useIndexTuple = IndexRelationGetNumberOfAttributes(index);
 
 	/*
 	 * Wait for index scans to complete. Scans before this point may contain
@@ -479,11 +477,8 @@ MarkDeleted(HnswVacuumState * vacuumstate)
 		/* Update element and neighbors together */
 		for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
 		{
-			ItemId		itemid = PageGetItemId(page, offno);
-			HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, itemid);
+			HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, PageGetItemId(page, offno));
 			HnswNeighborTuple ntup;
-			Size		etupSize;
-			Size		ntupSize;
 			Buffer		nbuf;
 			Page		npage;
 			BlockNumber neighborPage;
@@ -507,10 +502,6 @@ MarkDeleted(HnswVacuumState * vacuumstate)
 			if (ItemPointerIsValid(&etup->heaptids[0]))
 				continue;
 
-			/* Calculate sizes */
-			etupSize = ItemIdGetLength(itemid);
-			ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(etup->level, vacuumstate->m);
-
 			/* Get neighbor page */
 			neighborPage = ItemPointerGetBlockNumber(&etup->neighbortid);
 			neighborOffno = ItemPointerGetOffsetNumber(&etup->neighbortid);
@@ -531,19 +522,27 @@ MarkDeleted(HnswVacuumState * vacuumstate)
 
 			/* Overwrite element */
 			etup->deleted = 1;
-			MemSet(&etup->data, 0, VARSIZE_ANY(&etup->data));
+			if (useIndexTuple)
+			{
+				IndexTuple	itup = (IndexTuple) &etup->data;
+
+				MemSet(itup, 0, IndexTupleSize(itup));
+			}
+			else
+			{
+				Vector	   *vec = (Vector *) (&etup->data);
+
+				MemSet(vec, 0, VARSIZE_ANY(vec));
+			}
 
 			/* Overwrite neighbors */
 			for (int i = 0; i < ntup->count; i++)
 				ItemPointerSetInvalid(&ntup->indextids[i]);
 
-			/* Overwrite element tuple */
-			if (!PageIndexTupleOverwrite(page, offno, (Item) etup, etupSize))
-				elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
-
-			/* Overwrite neighbor tuple */
-			if (!PageIndexTupleOverwrite(npage, neighborOffno, (Item) ntup, ntupSize))
-				elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
+			/*
+			 * We modified the tuples in place, no need to call
+			 * PageIndexTupleOverwrite
+			 */
 
 			/* Commit */
 			GenericXLogFinish(state);
@@ -588,7 +587,7 @@ InitVacuumState(HnswVacuumState * vacuumstate, IndexVacuumInfo *info, IndexBulkD
 	vacuumstate->bas = GetAccessStrategy(BAS_BULKREAD);
 	vacuumstate->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
 	vacuumstate->collation = index->rd_indcollation[0];
-	vacuumstate->ntup = palloc0(BLCKSZ);
+	vacuumstate->ntup = palloc0(HNSW_TUPLE_ALLOC_SIZE);
 	vacuumstate->tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
 												"Hnsw vacuum temporary context",
 												ALLOCSET_DEFAULT_SIZES);

src/ivfbuild.c

@@ -29,8 +29,6 @@
 #define CALLBACK_ITEM_POINTER HeapTuple hup
 #endif
 
-#define UpdateProgress(index, val) pgstat_progress_update_param(index, val)
-
 #if PG_VERSION_NUM >= 140000
 #include "utils/backend_status.h"
 #include "utils/wait_event.h"
@@ -262,9 +260,9 @@ InsertTuples(Relation index, IvfflatBuildState * buildstate, ForkNumber forkNum)
 	TupleTableSlot *slot = MakeSingleTupleTableSlot(buildstate->tupdesc, &TTSOpsMinimalTuple);
 	TupleDesc	tupdesc = RelationGetDescr(index);
 
-	UpdateProgress(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_LOAD);
+	pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_LOAD);
 
-	UpdateProgress(PROGRESS_CREATEIDX_TUPLES_TOTAL, buildstate->indtuples);
+	pgstat_progress_update_param(PROGRESS_CREATEIDX_TUPLES_TOTAL, buildstate->indtuples);
 
 	GetNextTuple(buildstate->sortstate, tupdesc, slot, &itup, &list);
 
@@ -300,7 +298,7 @@ InsertTuples(Relation index, IvfflatBuildState * buildstate, ForkNumber forkNum)
 
 			pfree(itup);
 
-			UpdateProgress(PROGRESS_CREATEIDX_TUPLES_DONE, ++inserted);
+			pgstat_progress_update_param(PROGRESS_CREATEIDX_TUPLES_DONE, ++inserted);
 
 			GetNextTuple(buildstate->sortstate, tupdesc, slot, &itup, &list);
 		}
@@ -400,7 +398,7 @@ ComputeCenters(IvfflatBuildState * buildstate)
 {
 	int			numSamples;
 
-	UpdateProgress(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_KMEANS);
+	pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_KMEANS);
 
 	/* Target 50 samples per list, with at least 10000 samples */
 	/* The number of samples has a large effect on index build time */
@@ -475,7 +473,7 @@ CreateListPages(Relation index, VectorArray centers, int dimensions,
 	IvfflatList list;
 
 	listSize = MAXALIGN(IVFFLAT_LIST_SIZE(dimensions));
-	list = palloc(listSize);
+	list = palloc0(listSize);
 
 	buf = IvfflatNewBuffer(index, forkNum);
 	IvfflatInitRegisterPage(index, &buf, &page, &state);
@@ -921,7 +919,7 @@ AssignTuples(IvfflatBuildState * buildstate)
 	Oid			sortCollations[] = {InvalidOid};
 	bool		nullsFirstFlags[] = {false};
 
-	UpdateProgress(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_ASSIGN);
+	pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_ASSIGN);
 
 	/* Calculate parallel workers */
 	if (buildstate->heap != NULL)

src/ivfscan.c

@@ -310,6 +310,7 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
 		ItemPointer heaptid = (ItemPointer) DatumGetPointer(slot_getattr(so->slot, 2, &so->isnull));
 
 		scan->xs_heaptid = *heaptid;
+		scan->xs_recheck = false;
 		scan->xs_recheckorderby = false;
 		return true;
 	}

src/vector.c

@@ -10,7 +10,6 @@
 #include "lib/stringinfo.h"
 #include "libpq/pqformat.h"
 #include "port.h"				/* for strtof() */
-#include "port/pg_bitutils.h"
 #include "utils/array.h"
 #include "utils/builtins.h"
 #include "utils/float.h"
@@ -867,9 +866,10 @@ vector_mul(PG_FUNCTION_ARGS)
 int
 vector_cmp_internal(Vector * a, Vector * b)
 {
-	CheckDims(a, b);
+	int			dim = Min(a->dim, b->dim);
 
-	for (int i = 0; i < a->dim; i++)
+	/* Check values before dimensions to be consistent with Postgres arrays */
+	for (int i = 0; i < dim; i++)
 	{
 		if (a->x[i] < b->x[i])
 			return -1;
@@ -877,6 +877,13 @@ vector_cmp_internal(Vector * a, Vector * b)
 		if (a->x[i] > b->x[i])
 			return 1;
 	}
+
+	if (a->dim < b->dim)
+		return -1;
+
+	if (a->dim > b->dim)
+		return 1;
+
 	return 0;
 }
 
@@ -890,6 +897,9 @@ vector_lt(PG_FUNCTION_ARGS)
 	Vector	   *a = PG_GETARG_VECTOR_P(0);
 	Vector	   *b = PG_GETARG_VECTOR_P(1);
 
+	/* TODO Remove in 0.7.0 */
+	CheckDims(a, b);
+
 	PG_RETURN_BOOL(vector_cmp_internal(a, b) < 0);
 }
 
@@ -903,6 +913,9 @@ vector_le(PG_FUNCTION_ARGS)
 	Vector	   *a = PG_GETARG_VECTOR_P(0);
 	Vector	   *b = PG_GETARG_VECTOR_P(1);
 
+	/* TODO Remove in 0.7.0 */
+	CheckDims(a, b);
+
 	PG_RETURN_BOOL(vector_cmp_internal(a, b) <= 0);
 }
 
@@ -916,6 +929,9 @@ vector_eq(PG_FUNCTION_ARGS)
 	Vector	   *a = PG_GETARG_VECTOR_P(0);
 	Vector	   *b = PG_GETARG_VECTOR_P(1);
 
+	/* TODO Remove in 0.7.0 */
+	CheckDims(a, b);
+
 	PG_RETURN_BOOL(vector_cmp_internal(a, b) == 0);
 }
 
@@ -929,6 +945,9 @@ vector_ne(PG_FUNCTION_ARGS)
 	Vector	   *a = PG_GETARG_VECTOR_P(0);
 	Vector	   *b = PG_GETARG_VECTOR_P(1);
 
+	/* TODO Remove in 0.7.0 */
+	CheckDims(a, b);
+
 	PG_RETURN_BOOL(vector_cmp_internal(a, b) != 0);
 }
 
@@ -942,6 +961,9 @@ vector_ge(PG_FUNCTION_ARGS)
 	Vector	   *a = PG_GETARG_VECTOR_P(0);
 	Vector	   *b = PG_GETARG_VECTOR_P(1);
 
+	/* TODO Remove in 0.7.0 */
+	CheckDims(a, b);
+
 	PG_RETURN_BOOL(vector_cmp_internal(a, b) >= 0);
 }
 
@@ -955,6 +977,9 @@ vector_gt(PG_FUNCTION_ARGS)
 	Vector	   *a = PG_GETARG_VECTOR_P(0);
 	Vector	   *b = PG_GETARG_VECTOR_P(1);
 
+	/* TODO Remove in 0.7.0 */
+	CheckDims(a, b);
+
 	PG_RETURN_BOOL(vector_cmp_internal(a, b) > 0);
 }
 
@@ -1135,43 +1160,3 @@ vector_avg(PG_FUNCTION_ARGS)
 
 	PG_RETURN_POINTER(result);
 }
-
-/*
- * Ensure same number of bytes
- */
-static inline void
-CheckByteLengths(uint32 aLen, uint32 bLen)
-{
-	if (aLen != bLen)
-		ereport(ERROR,
-				(errcode(ERRCODE_DATA_EXCEPTION),
-				 errmsg("different byte lengths %u and %u", aLen, bLen)));
-}
-
-/*
- * Get the Hamming distance between two binary strings
- */
-PGDLLEXPORT PG_FUNCTION_INFO_V1(hamming_distance);
-Datum
-hamming_distance(PG_FUNCTION_ARGS)
-{
-	bytea	   *a = PG_GETARG_BYTEA_PP(0);
-	bytea	   *b = PG_GETARG_BYTEA_PP(1);
-	char	   *ax = VARDATA_ANY(a);
-	char	   *bx = VARDATA_ANY(b);
-	uint32		aLen = VARSIZE_ANY_EXHDR(a);
-	uint32		bLen = VARSIZE_ANY_EXHDR(b);
-	uint64		distance = 0;
-
-	CheckByteLengths(aLen, bLen);
-
-	for (uint32 i = 0; i < aLen; i++)
-	{
-		unsigned char diff = (unsigned char) (ax[i] ^ bx[i]);
-
-		distance += pg_number_of_ones[diff];
-	}
-
-	/* TODO Decide on return type */
-	PG_RETURN_FLOAT8((double) distance);
-}

test/expected/functions.out

@@ -24,6 +24,56 @@ SELECT '[1e37]'::vector * '[1e37]';
 ERROR:  value out of range: overflow
 SELECT '[1e-37]'::vector * '[1e-37]';
 ERROR:  value out of range: underflow
+SELECT '[1,2,3]'::vector = '[1,2,3]';
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT '[1,2,3]'::vector = '[1,2]';
+ERROR:  different vector dimensions 3 and 2
+SELECT vector_cmp('[1,2,3]', '[1,2,3]');
+ vector_cmp 
+------------
+          0
+(1 row)
+
+SELECT vector_cmp('[1,2,3]', '[0,0,0]');
+ vector_cmp 
+------------
+          1
+(1 row)
+
+SELECT vector_cmp('[0,0,0]', '[1,2,3]');
+ vector_cmp 
+------------
+         -1
+(1 row)
+
+SELECT vector_cmp('[1,2]', '[1,2,3]');
+ vector_cmp 
+------------
+         -1
+(1 row)
+
+SELECT vector_cmp('[1,2,3]', '[1,2]');
+ vector_cmp 
+------------
+          1
+(1 row)
+
+SELECT vector_cmp('[1,2]', '[2,3,4]');
+ vector_cmp 
+------------
+         -1
+(1 row)
+
+SELECT vector_cmp('[2,3]', '[1,2,3]');
+ vector_cmp 
+------------
+          1
+(1 row)
+
 SELECT vector_dims('[1,2,3]');
  vector_dims 
 -------------
@@ -158,32 +208,6 @@ SELECT l1_distance('[3e38]', '[-3e38]');
     Infinity
 (1 row)
 
-SELECT hamming_distance('\xFFFF', '\xFFFF');
- hamming_distance 
-------------------
-                0
-(1 row)
-
-SELECT hamming_distance('\xFFFF', '\xFFFE');
- hamming_distance 
-------------------
-                1
-(1 row)
-
-SELECT hamming_distance('\xFFFF', '\xFFFC');
- hamming_distance 
-------------------
-                2
-(1 row)
-
-SELECT hamming_distance('\xFFFF', '\x0000');
- hamming_distance 
-------------------
-               16
-(1 row)
-
-SELECT hamming_distance('\xFFFF', '\x00');
-ERROR:  different byte lengths 2 and 1
 SELECT avg(v) FROM unnest(ARRAY['[1,2,3]'::vector, '[3,5,7]']) v;
     avg    
 -----------

test/sql/functions.sql

@@ -6,6 +6,17 @@ SELECT '[1,2,3]'::vector * '[4,5,6]';
 SELECT '[1e37]'::vector * '[1e37]';
 SELECT '[1e-37]'::vector * '[1e-37]';
 
+SELECT '[1,2,3]'::vector = '[1,2,3]';
+SELECT '[1,2,3]'::vector = '[1,2]';
+
+SELECT vector_cmp('[1,2,3]', '[1,2,3]');
+SELECT vector_cmp('[1,2,3]', '[0,0,0]');
+SELECT vector_cmp('[0,0,0]', '[1,2,3]');
+SELECT vector_cmp('[1,2]', '[1,2,3]');
+SELECT vector_cmp('[1,2,3]', '[1,2]');
+SELECT vector_cmp('[1,2]', '[2,3,4]');
+SELECT vector_cmp('[2,3]', '[1,2,3]');
+
 SELECT vector_dims('[1,2,3]');
 
 SELECT round(vector_norm('[1,1]')::numeric, 5);
@@ -37,12 +48,6 @@ SELECT l1_distance('[0,0]', '[0,1]');
 SELECT l1_distance('[1,2]', '[3]');
 SELECT l1_distance('[3e38]', '[-3e38]');
 
-SELECT hamming_distance('\xFFFF', '\xFFFF');
-SELECT hamming_distance('\xFFFF', '\xFFFE');
-SELECT hamming_distance('\xFFFF', '\xFFFC');
-SELECT hamming_distance('\xFFFF', '\x0000');
-SELECT hamming_distance('\xFFFF', '\x00');
-
 SELECT avg(v) FROM unnest(ARRAY['[1,2,3]'::vector, '[3,5,7]']) v;
 SELECT avg(v) FROM unnest(ARRAY['[1,2,3]'::vector, '[3,5,7]', NULL]) v;
 SELECT avg(v) FROM unnest(ARRAY[]::vector[]) v;

test/t/020_hnsw_filtering.pl

@@ -0,0 +1,154 @@
+use strict;
+use warnings;
+use PostgresNode;
+use TestLib;
+use Test::More;
+
+my $node;
+my @queries = ();
+my @cs = ();
+my @expected;
+my $limit = 20;
+my $dim = 3;
+my $array_sql = join(",", ('random()') x $dim);
+my $nc = 50;
+my @types = ("int4", "int8", "text", "uuid", "varchar");
+my $type = $types[rand(@types)];
+
+sub cast_c
+{
+	my ($c) = @_;
+
+	if ($type eq "uuid")
+	{
+		return "('00000000-0000-0000-0000-0000000000' || LPAD(($c)::text, 2, '0'))::uuid";
+	}
+	else
+	{
+		return "($c)::$type";
+	}
+}
+
+sub test_recall
+{
+	my ($min, $operator) = @_;
+	my $correct = 0;
+	my $total = 0;
+
+	my $explain = $node->safe_psql("postgres", qq(
+		SET enable_seqscan = off;
+		EXPLAIN ANALYZE SELECT i FROM tst WHERE c = @{[cast_c($cs[0])]} ORDER BY v $operator '$queries[0]' LIMIT $limit;
+	));
+	like($explain, qr/Index Cond/);
+
+	for my $i (0 .. $#queries)
+	{
+		my $actual = $node->safe_psql("postgres", qq(
+			SET enable_seqscan = off;
+			SELECT i FROM tst WHERE c = @{[cast_c($cs[$i])]} ORDER BY v $operator '$queries[$i]' LIMIT $limit;
+		));
+		my @actual_ids = split("\n", $actual);
+		my %actual_set = map { $_ => 1 } @actual_ids;
+
+		is(scalar(@actual_ids), $limit);
+
+		my @expected_ids = split("\n", $expected[$i]);
+
+		foreach (@expected_ids)
+		{
+			if (exists($actual_set{$_}))
+			{
+				$correct++;
+			}
+			$total++;
+		}
+	}
+
+	cmp_ok($correct / $total, ">=", $min, $operator);
+}
+
+# Initialize node
+$node = get_new_node('node');
+$node->init;
+$node->start;
+
+# Create table
+$node->safe_psql("postgres", "CREATE EXTENSION vector;");
+$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim), c $type, c2 $type);");
+$node->safe_psql("postgres", qq(
+	INSERT INTO tst SELECT i, ARRAY[$array_sql], @{[cast_c("i % $nc")]}, @{[cast_c("i % $nc")]} FROM generate_series(1, 10000) i;
+));
+$node->safe_psql("postgres", "CREATE INDEX ON tst USING hnsw (v vector_l2_ops, c, c2);");
+$node->safe_psql("postgres", qq(
+	INSERT INTO tst SELECT i, ARRAY[$array_sql], @{[cast_c("i % $nc")]}, @{[cast_c("i % $nc")]} FROM generate_series(1, 10000) i;
+));
+$node->safe_psql("postgres", "ANALYZE tst;");
+
+# Generate queries
+for (1 .. 20)
+{
+	my @r = ();
+	for (1 .. $dim)
+	{
+		push(@r, rand());
+	}
+	push(@queries, "[" . join(",", @r) . "]");
+	push(@cs, int(rand() * $nc));
+}
+
+# Get exact results
+@expected = ();
+for my $i (0 .. $#queries)
+{
+	my $res = $node->safe_psql("postgres", qq(
+		SET enable_indexscan = off;
+		SELECT i FROM tst WHERE c = @{[cast_c($cs[$i])]} ORDER BY v <-> '$queries[$i]' LIMIT $limit;
+	));
+	push(@expected, $res);
+}
+
+# Test recall
+test_recall(0.99, '<->');
+
+# Test no conditions
+my $explain = $node->safe_psql("postgres", qq(
+	EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v <-> '$queries[0]' LIMIT $limit;
+));
+like($explain, qr/Index Scan/);
+
+# Test range
+$explain = $node->safe_psql("postgres", qq(
+	EXPLAIN ANALYZE SELECT i FROM tst WHERE c >= @{[cast_c(1)]} AND c <= @{[cast_c(3)]} ORDER BY v <-> '$queries[0]' LIMIT $limit;
+));
+like($explain, qr/Index Cond: \(\(\S+ >= \S+\) AND \(\S+ <= \S+\)\)/);
+
+# Test multiple conditions
+$explain = $node->safe_psql("postgres", qq(
+	EXPLAIN ANALYZE SELECT i FROM tst WHERE c = @{[cast_c($cs[0])]} AND c2 = @{[cast_c($cs[0])]} ORDER BY v <-> '$queries[0]' LIMIT $limit;
+));
+like($explain, qr/Index Cond: \(\(\S+ = \S+\) AND \(\S+ = \S+\)\)/);
+
+# Test no order
+$explain = $node->safe_psql("postgres", qq(
+	EXPLAIN ANALYZE SELECT i FROM tst WHERE c = @{[cast_c($cs[0])]} LIMIT $limit;
+));
+like($explain, qr/Seq Scan/);
+
+# Test vacuum
+$node->safe_psql("postgres", "DELETE FROM tst WHERE c > @{[cast_c(5)]};");
+$node->safe_psql("postgres", "VACUUM tst;");
+
+# Test columns
+my ($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING hnsw (c);");
+like($stderr, qr/first column must be a vector/);
+
+($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING hnsw (c, v vector_l2_ops);");
+like($stderr, qr/first column must be a vector/);
+
+($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING hnsw (v vector_l2_ops, c, c, c);");
+like($stderr, qr/index cannot have more than three columns/);
+
+($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING hnsw (v vector_l2_ops, v vector_l2_ops);");
+like($stderr, qr/column 2 cannot be a vector/);
+
+done_testing();

vector.control

@@ -1,4 +1,4 @@
 comment = 'vector data type and ivfflat and hnsw access methods'
-default_version = '0.5.1'
+default_version = '0.6.0'
 module_pathname = '$libdir/vector'
 relocatable = true