Lantern v0.0.10

What's Changed

• Clarify operator modes in README by @dqii in #242
• WAL/blockmaps: continue blockmaps initialization after failure by @var77 in #241
• Add WAL records for all neighbor modifications during insert by @medvied in #215
• Fix l2 operator bug when used with int arrays by @Ngalstyan4 in #243
• Rename generic operator. by @var77 in #244

Full Changelog: v0.0.9...v0.0.10

Lantern v0.0.9

What's Changed

• Fix memory context for List node allocation in op_rewrite.c by @var77 in #234
• Add C client tests to catch op_rewriting segfault by @var77 in #238
• Added dockerfile with postgres debug build by @var77 in #233
• Add flag to disable operator rewriting hooks and make pgvector compatible by @var77 in #240

Full Changelog: v0.0.8...v0.0.9

Lantern v0.0.8

What's Changed

• Narek/release0.0.8 by @Ngalstyan4 in #235

Full Changelog: v0.0.7...v0.0.8

Lantern v0.0.7

What's Changed

• Fix wrong memory context for allocation in op_rewrite.c by @var77 in #231
• Release 0.0.7 by @Ngalstyan4 in #232

Full Changelog: v0.0.6...v0.0.7

Lantern v0.0.6

What's Changed

• Add support for <-> operator rewriting by @ezra-varady in #189
• Fix missing handling for operator usage in materialized queries by @Ngalstyan4 in #228

Full Changelog: v0.0.5...v0.0.6

Lantern v0.0.5

What's Changed

• Fix package build regression. Make sure built packages are portible by @Ngalstyan4 in #138
• Update docker run instructions by @Ngalstyan4 in #139
• fix: dim not dims by @dqii in #141
• Add seqscan to README by @dqii in #142
• Add cloud waitlist by @dqii in #154
• Fix: Add Lantern database URL to support new benchmarking setup by @dqii in #156
• Support Postgres 16 by @dqii in #158
• Add Mac builds into CI pipeline by @var77 in #143
• Set module extension to dylib for mac and postgres 16 by @var77 in #165
• Check for undefined symbols #145 by @var77 in #155
• Create source code archive manually by @var77 in #173
• Misc changes from several PRs (look at individual commits) by @Ngalstyan4 in #176
• Replace fa_cache with htable cache by @var77 in #160
• Make check_symbols not accept dynamic c++ symbols by @Ngalstyan4 in #181
• Support expression-based index operator checks by @dqii in #179
• Homebrew formula release by @var77 in #183
• Infer array dimensions from index expressions by @var77 in #175
• Fix operator classes for lantern_hnsw access method by @var77 in #180
• Fix int overflow issue for progress by @var77 in #186
• Remove pgvector from Docker by @dqii in #184
• Run benchmarks on self-hosted runner by @var77 in #190
• Add parallel tests by @ezra-varady in #192
• Integer array type conversion in array_dist to compute hamming distance by @therealdarkknight in #191
• Revert to fa_cache for node retriever by @var77 in #203
• Add database version updates by @Ngalstyan4 in #193
• Narek/ezra work mem by @Ngalstyan4 in #205
• Update to BSL license by @dqii in #207
• Add support and tests for install-time schema relocation by @Ngalstyan4 in #206
• Added hnsw.ef_search variable to change expansion factor during search at runtime by @therealdarkknight in #199
• Figuring out why tests fail on main by @Ngalstyan4 in #214
• add action to run against asan by @ezra-varady in #212
• validate-index: implement a function to validate index data structures by @medvied in #208
• cmake/FindPostgreSQL: put the server include dir first by @medvied in #217
• fix incorrect minor version for postgres 13 by @ezra-varady in #216
• fix sanitizers not triggering properly by @ezra-varady in #220
• Rename hnsw.init_k to lantern_hnsw.init_k by @var77 in #219
• Implement failure points by @medvied in #218
• Update images in README and spacing edits by @dqii in #225
• Release Lantern v0.0.5 by @Ngalstyan4 in #223

New Contributors

• @medvied made their first contribution in #208

Full Changelog: v0.0.4...v0.0.5

Lantern v0.0.4

What's Changed

• Ndebug by @Ngalstyan4 in #31
• create index with dynamically allocated blockmaps by @Ngalstyan4 in #32
• [FEATURE] Avoid spilling index into file by @var77 in #35
• insert into empty index by @davkhech in #41
• Enable default cost estimator before we craft a usearch-hnsw specific one by @Ngalstyan4 in #36
• Ignore whitespace when running tests by @cppietime in #44
• Create CODE_OF_CONDUCT.md by @dqii in #47
• Docker CI Improvements by @var77 in #49
• Run tests with pg_regress by @var77 in #50
• Added autogeneration for lanterdb.control file from CMake by @grubdragon in #53
• Added dist functions from usearch by @var77 in #46
• Fix a double-free bug, triggered with 10k rows by @Ngalstyan4 in #61
• No global vars by @Ngalstyan4 in #51
• Fix readme and get rid of an extra import by @Ngalstyan4 in #65
• Add lanterndb_extras package into release by @var77 in #66
• Fix example in README by @var77 in #69
• Get rid of insert cache context for now to simplify global lock story by @Ngalstyan4 in #68
• Main dev by @Ngalstyan4 in #73
• Switch wget -> curl for downloading vector files for tests by @Ngalstyan4 in #75
• Dev workflow improvements by @dqii in #79
• Main dev by @Ngalstyan4 in #80
• Enable dynamic undefined symbols in shared libraries on Apple by @Ngalstyan4 in #76
• cleanup retrievers by @davkhech in #70
• Fix metric_kind param when doing index scan by @var77 in #82
• remove hard limit on the number of items returned when an index is queried by @ezra-varady in #78
• Enable compiler warnings and address most of them by @Ngalstyan4 in #86
• Test improvements by @dqii in #87
• Fix range of HNSW's M parameter to be >= 2, add tests by @dqii in #95
• Feature/ezra infer dim by @Ngalstyan4 in #98
• Add a fully associative FIFO cache for recently retrieved nodes by @Ngalstyan4 in #91
• Only run pgvector test if pgvector is installed by @dqii in #97
• Add ability to load index from file by @var77 in #99
• Add benchmarking to local dev and to CI/CD by @dqii in #100
• Hotfix: Fix Dockerfile.dev given benchmarking changes by @dqii in #102
• replace fixed size takenbuffer with linked list (#96) by @dqii in #104
• Incorporate HNSW-specific information to cost estimate by @yoloVoe in #103
• cleanup: get rid of unnecessary markbufferdirty and field by @yoloVoe in #106
• Fix ON_ERROR_STOP in a test by @Ngalstyan4 in #105
• Detect correct operator usage using AST by @dqii in #111
• Handle additional AST cases for operator usage checks by @dqii in #118
• Fix test runner for OS X, fix checking pgvector in tests by @var77 in #119
• Cleanup the root folder of the repository by @Ngalstyan4 in #121
• Save and load usearch options from index header page by @var77 in #101
• Clang format not correctly applied to C files by @dqii in #115
• Enable codecov check only for prs by @var77 in #123
• Fix missing benchmarking env variables by @dqii in #126
• Dynamically scale usearch index on construction by @ezra-varady in #122
• Add benchmark comment by @dqii in #127
• Fix benchmarking on push event by @var77 in #131
• Fix: Don't run benchmark comment for forks by @dqii in #132
• Throw error when index operator is used without index by @dqii in #124
• Cleanup and asserts by @Ngalstyan4 in #130
• Add pgvector 0.5.0+ vector type compatibility by @Ngalstyan4 in #128
• Rename dims->dim in index creation by @Ngalstyan4 in #129
• Reduce elog overhead on fastpath by @Ngalstyan4 in #120
• README updates by @dqii in #134
• Move file access check to build phase by @var77 in #136
• Rename lanterndb->lantern by @Ngalstyan4 in #137
• Release 0.0.4 by @Ngalstyan4 in #135

New Contributors

• @davkhech made their first contribution in #41
• @cppietime made their first contribution in #44
• @yoloVoe made their first contribution in #103

Full Changelog: v0.0.3...v0.0.4

LanternDB v0.0.3

What's Changed

• [FEATURE]: added codecov badge by @var77 in #17
• [FEATURE]: support real[] array type for hnsw index by @var77 in #18
• [FEATURE]: CI improvements, codecov comments, formatting check, github releases by @var77 in #23
• Typefix and livedebug by @Ngalstyan4 in #24
• [FEATURE] Change array cache to HTAB cache for wal_block_numbers by @var77 in #26
• Fix cache key by @var77 in #28
• Upgrade usearch by @Ngalstyan4 in #27
• Release 0.0.3 by @Ngalstyan4 in #29

If you are working on the source, please make sure you do not have changes on main branches of lanterndb and usearch and run:

git pull
git submodule update

and then, from postgres

ALTER extension lanterndb UPDATE;

Note: upgrades as described above are only supported if you were running LanternDB 0.0.2. Please drop and recreate LanternDB postgres extension if you are running an older version.
Alternatively, you can run git clone --recursive [email protected]:lanterndata/lanterndb.git

Full Changelog: v0.0.0...v0.0.3

LanternDB v0.0.2

What's Changed

• Implement inserts into an indexed table by @Ngalstyan4 in #1
• [FEATURE]: CI script to build and run tests on 11-15 versions of PG by @var77 in #2
• [FEATURE]: Dockerfile and Docker build action by @var77 in #5
• Update README.md by @Ngalstyan4 in #8
• Optimizations on inserts by @Ngalstyan4 in #3
• Fix CI tests by @Ngalstyan4 in #9
• [FEATURE]: Added debian and archive releases by @var77 in #15
• [FEATURE]: separate workflow files, add test coverage by @var77 in #16

New Contributors

• @Ngalstyan4 made their first contribution in #1

Full Changelog: https://github.com/lanterndata/lanterndb/commits/v0.0.2