Version bump to 0.4.2 [skip ci]

Presently, array_to_vector()'s call to CheckDim() is skipped when typmod != -1, which allows for bypassing VECTOR_MAX_DIM.  To fix, call Check[Expected]Dim() unconditionally.  CheckExpectedDim() takes no action when typmod == -1, so there's no need to guard it with an 'if' statement.

CHANGELOG.md

@@ -1,6 +1,8 @@
-## 0.4.2 (unreleased)
+## 0.4.2 (2023-05-13)
 
 - Added notice when index created with little data
+- Fixed dimensions check for some direct function calls
+- Fixed installation error with Postgres 12.0-12.2
 
 ## 0.4.1 (2023-03-21)
 

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": "0.4.1",
+	"version": "0.4.2",
 	"maintainer": [
 		"Andrew Kane <andrew@ankane.org>"
 	],
@@ -20,7 +20,7 @@
 		"vector": {
 			"file": "sql/vector.sql",
 			"docfile": "README.md",
-			"version": "0.4.1",
+			"version": "0.4.2",
 			"abstract": "Open-source vector similarity search for Postgres"
 		}
 	},

Makefile

@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.4.1
+EXTVERSION = 0.4.2
 
 MODULE_big = vector
 DATA = $(wildcard sql/*--*.sql)
@@ -63,3 +63,9 @@ dist:
 
 docker:
 	docker build --pull --no-cache --platform linux/amd64 -t ankane/pgvector:latest .
+
+.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) .

Makefile.win

@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.4.1
+EXTVERSION = 0.4.2
 
 OBJS = src\ivfbuild.obj src\ivfflat.obj src\ivfinsert.obj src\ivfkmeans.obj src\ivfscan.obj src\ivfutils.obj src\ivfvacuum.obj src\vector.obj
 

README.md

@@ -2,7 +2,11 @@
 
 Open-source vector similarity search for Postgres
 
-Supports exact and approximate nearest neighbor search for L2 distance, inner product, and cosine distance
+Supports
+
+- exact and approximate nearest neighbor search
+- L2 distance, inner product, and cosine distance
+- any [language](#languages) with a Postgres client
 
 [![Build Status](https://github.com/pgvector/pgvector/workflows/build/badge.svg?branch=master)](https://github.com/pgvector/pgvector/actions)
 
@@ -12,7 +16,7 @@ Compile and install the extension (supports Postgres 11+)
 
 ```sh
 cd /tmp
-git clone --branch v0.4.1 https://github.com/pgvector/pgvector.git
+git clone --branch v0.4.2 https://github.com/pgvector/pgvector.git
 cd pgvector
 make
 make install # may need sudo
@@ -36,7 +40,7 @@ Create a vector column with 3 dimensions
 CREATE TABLE items (id bigserial PRIMARY KEY, embedding vector(3));
 ```
 
-Insert values
+Insert vectors
 
 ```sql
 INSERT INTO items (embedding) VALUES ('[1,2,3]'), ('[4,5,6]');
@@ -124,7 +128,7 @@ SELECT embedding <-> '[3,1,2]' AS distance FROM items;
 For inner product, multiply by -1 (since `<#>` returns the negative inner product)
 
 ```sql
-SELECT -1 * (embedding <#> '[3,1,2]') AS inner_product FROM items;
+SELECT (embedding <#> '[3,1,2]') * -1 AS inner_product FROM items;
 ```
 
 For cosine similarity, use 1 - cosine distance
@@ -133,7 +137,7 @@ For cosine similarity, use 1 - cosine distance
 SELECT 1 - (embedding <=> '[3,1,2]') AS cosine_similarity FROM items;
 ```
 
-#### Averaging
+#### Aggregates
 
 Average vectors
 
@@ -156,8 +160,8 @@ You can add an index to use approximate nearest neighbor search, which trades so
 Three keys to achieving good recall are:
 
 1. Create the index *after* the table has some data
-2. Choose an appropriate number of lists (a good place to start is `rows / 1000` for up to 1M rows and `sqrt(rows)` for over 1M rows)
+2. Choose an appropriate number of lists - a good place to start is `rows / 1000` for up to 1M rows and `sqrt(rows)` for over 1M rows
-3. When querying, specify an appropriate number of [probes](#query-options) (higher is better for recall, lower is better for speed)
+3. When querying, specify an appropriate number of [probes](#query-options) (higher is better for recall, lower is better for speed) - a good place to start is `lists / 10` for up to 1M rows and `sqrt(lists)` for over 1M rows
 
 Add an index for each distance function you want to use.
 
@@ -275,8 +279,10 @@ Language | Libraries / Examples
 --- | ---
 C++ | [pgvector-cpp](https://github.com/pgvector/pgvector-cpp)
 C# | [pgvector-dotnet](https://github.com/pgvector/pgvector-dotnet)
+Crystal | [pgvector-crystal](https://github.com/pgvector/pgvector-crystal)
 Elixir | [pgvector-elixir](https://github.com/pgvector/pgvector-elixir)
 Go | [pgvector-go](https://github.com/pgvector/pgvector-go)
+Haskell | [pgvector-haskell](https://github.com/pgvector/pgvector-haskell)
 Java, Scala | [pgvector-java](https://github.com/pgvector/pgvector-java)
 Julia | [pgvector-julia](https://github.com/pgvector/pgvector-julia)
 Lua | [pgvector-lua](https://github.com/pgvector/pgvector-lua)
@@ -287,6 +293,7 @@ Python | [pgvector-python](https://github.com/pgvector/pgvector-python)
 R | [pgvector-r](https://github.com/pgvector/pgvector-r)
 Ruby | [pgvector-ruby](https://github.com/pgvector/pgvector-ruby), [Neighbor](https://github.com/ankane/neighbor)
 Rust | [pgvector-rust](https://github.com/pgvector/pgvector-rust)
+Swift | [pgvector-swift](https://github.com/pgvector/pgvector-swift)
 
 ## Frequently Asked Questions
 
@@ -300,10 +307,11 @@ Yes, pgvector uses the write-ahead log (WAL), which allows for replication and p
 
 #### What if I want to index vectors with more than 2,000 dimensions?
 
-Two things you can try are:
+You’ll need to use [dimensionality reduction](https://en.wikipedia.org/wiki/Dimensionality_reduction) at the moment.
 
-1. use dimensionality reduction
+#### Why am I seeing less results after adding an index?
-2. compile Postgres with a larger block size (`./configure --with-blocksize=32`) and edit the limit in `src/ivfflat.h`
+
+The index was likely created with too little data for the number of lists. Drop the index until the table has more data.
 
 ## Reference
 
@@ -347,7 +355,11 @@ If your machine has multiple Postgres installations, specify the path to [pg_con
 export PG_CONFIG=/Applications/Postgres.app/Contents/Versions/latest/bin/pg_config
 ```
 
-Then re-run the installation instructions (run `make clean` before `make` if needed)
+Then re-run the installation instructions (run `make clean` before `make` if needed). If `sudo` is needed for `make install`, use:
+
+```sh
+sudo --preserve-env=PG_CONFIG make install
+```
 
 ### Missing Header
 
@@ -367,7 +379,7 @@ Support for Windows is currently experimental. Use `nmake` to build:
 
 ```cmd
 set "PGROOT=C:\Program Files\PostgreSQL\15"
-git clone --branch v0.4.1 https://github.com/pgvector/pgvector.git
+git clone --branch v0.4.2 https://github.com/pgvector/pgvector.git
 cd pgvector
 nmake /F Makefile.win
 nmake /F Makefile.win install
@@ -388,7 +400,7 @@ This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (r
 You can also build the image manually:
 
 ```sh
-git clone --branch v0.4.1 https://github.com/pgvector/pgvector.git
+git clone --branch v0.4.2 https://github.com/pgvector/pgvector.git
 cd pgvector
 docker build -t pgvector .
 ```
@@ -401,6 +413,8 @@ With Homebrew Postgres, you can use:
 brew install pgvector
 ```
 
+Note: This only adds it to the `postgresql@14` formula
+
 ### PGXN
 
 Install from the [PostgreSQL Extension Network](https://pgxn.org/dist/vector) with:
@@ -437,11 +451,10 @@ pgvector is available on [these providers](https://github.com/pgvector/pgvector/
 
 To request a new extension on other providers:
 
-- Amazon RDS - follow the instructions on [this page](https://aws.amazon.com/rds/postgresql/faqs/)
 - Google Cloud SQL - vote or comment on [this page](https://issuetracker.google.com/issues/265172065)
 - Azure Database - vote or comment on [this page](https://feedback.azure.com/d365community/idea/7b423322-6189-ed11-a81b-000d3ae49307)
-- DigitalOcean Managed Databases - vote or comment on [this page](https://ideas.digitalocean.com/app-framework-services/p/pgvector-extension-for-postgresql)
+- DigitalOcean Managed Databases - vote or comment on [this page](https://ideas.digitalocean.com/managed-database/p/pgvector-extension-for-postgresql)
-- Render - vote or comment on [this page](https://feedback.render.com/features/p/add-pgvector-extension-to-postgresql)
+- Heroku Postgres - vote or comment on [this page](https://github.com/heroku/roadmap/issues/156)
 
 ## Upgrading
 

sql/vector--0.4.1--0.4.2.sql

@@ -0,0 +1,2 @@
+-- complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "ALTER EXTENSION vector UPDATE TO '0.4.2'" to load this file. \quit

src/ivfbuild.c

@@ -438,8 +438,8 @@ ComputeCenters(IvfflatBuildState * buildstate)
 		{
 			ereport(NOTICE,
 					(errmsg("ivfflat index created with little data"),
-					 errdetail("this will cause poor recall"),
+					 errdetail("This will cause low recall."),
-					 errhint("drop the index until the table has more data")));
+					 errhint("Drop the index until the table has more data.")));
 		}
 	}
 

src/ivfscan.c

@@ -124,7 +124,7 @@ GetScanItems(IndexScanDesc scan, Datum value)
 	 *
 	 * See postgres/src/backend/storage/buffer/README for description
 	 */
-	BufferAccessStrategy bas = GetAccessStrategy(BAS_NORMAL);
+	BufferAccessStrategy bas = GetAccessStrategy(BAS_BULKREAD);
 
 	/* Search closest probes lists */
 	while (!pairingheap_is_empty(so->listQueue))
@@ -174,8 +174,8 @@ GetScanItems(IndexScanDesc scan, Datum value)
 	if (tuples < 100)
 		ereport(DEBUG1,
 				(errmsg("index scan found few tuples"),
-				 errdetail("index may have been created without data or lists is too high"),
+				 errdetail("Index may have been created with little data."),
-				 errhint("recreate the index and possibly decrease lists")));
+				 errhint("Recreate the index and possibly decrease lists.")));
 
 	tuplesort_performsort(so->sortstate);
 }

src/vector.c

@@ -100,7 +100,7 @@ CheckStateArray(ArrayType *statearray, const char *caller)
 	return (float8 *) ARR_DATA_PTR(statearray);
 }
 
-#if PG_VERSION_NUM < 120000
+#if PG_VERSION_NUM < 120003
 static pg_noinline void
 float_overflow_error(void)
 {
@@ -396,10 +396,8 @@ array_to_vector(PG_FUNCTION_ARGS)
 	get_typlenbyvalalign(ARR_ELEMTYPE(array), &typlen, &typbyval, &typalign);
 	deconstruct_array(array, ARR_ELEMTYPE(array), typlen, typbyval, typalign, &elemsp, &nullsp, &nelemsp);
 
-	if (typmod == -1)
+	CheckDim(nelemsp);
-		CheckDim(nelemsp);
+	CheckExpectedDim(typmod, nelemsp);
-	else
-		CheckExpectedDim(typmod, nelemsp);
 
 	result = InitVector(nelemsp);
 	for (i = 0; i < nelemsp; i++)
@@ -952,6 +950,7 @@ vector_avg(PG_FUNCTION_ARGS)
 
 	/* Create vector */
 	dim = STATE_DIMS(statearray);
+	CheckDim(dim);
 	result = InitVector(dim);
 	for (int i = 0; i < dim; i++)
 	{

test/expected/cast.out

@@ -46,6 +46,8 @@ SELECT '[1,2,3]'::vector::real[];
 
 SELECT array_agg(n)::vector FROM generate_series(1, 16001) n;
 ERROR:  vector cannot have more than 16000 dimensions
+SELECT array_to_vector(array_agg(n), 16001, false) FROM generate_series(1, 16001) n;
+ERROR:  vector cannot have more than 16000 dimensions
 -- ensure no error
 SELECT ARRAY[1,2,3] = ARRAY[1,2,3];
  ?column? 

test/expected/functions.out

@@ -102,3 +102,5 @@ SELECT avg(v) FROM unnest(ARRAY[]::vector[]) v;
 
 SELECT avg(v) FROM unnest(ARRAY['[1,2]'::vector, '[3]']) v;
 ERROR:  expected 2 dimensions, not 1
+SELECT vector_avg(array_agg(n)) FROM generate_series(1, 16002) n;
+ERROR:  vector cannot have more than 16000 dimensions

test/sql/cast.sql

@@ -10,6 +10,7 @@ SELECT '{-Infinity}'::real[]::vector;
 SELECT '{}'::real[]::vector;
 SELECT '[1,2,3]'::vector::real[];
 SELECT array_agg(n)::vector FROM generate_series(1, 16001) n;
+SELECT array_to_vector(array_agg(n), 16001, false) FROM generate_series(1, 16001) n;
 
 -- ensure no error
 SELECT ARRAY[1,2,3] = ARRAY[1,2,3];

test/sql/functions.sql

@@ -24,3 +24,4 @@ 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;
 SELECT avg(v) FROM unnest(ARRAY['[1,2]'::vector, '[3]']) v;
+SELECT vector_avg(array_agg(n)) FROM generate_series(1, 16002) n;

vector.control

@@ -1,4 +1,4 @@
 comment = 'vector data type and ivfflat access method'
-default_version = '0.4.1'
+default_version = '0.4.2'
 module_pathname = '$libdir/vector'
 relocatable = true