v0.1.0

What's Changed

• Add bench harness for fts and for vector search by @vinaykakade in #1
• test: rename id-column rejection test, broaden coverage by @vinaykakade in #2
• fix benchmarks: make benchmarks runnable using criterion by @Pranav2612000 in #3
• chore: add rustfmt.toml for uniform formatting by @Pranav2612000 in #4
• benchmarks: allow running superfile and supertable benchmarks seperately by @Pranav2612000 in #6
• Add peak-RSS stats to benchmarks by @ekechinwokah in #7
• bench: report RSS median and p90 by @ekechinwokah in #8
• Accelerate CRC verification to improve cold open latency by @ekechinwokah in #9
• bench: suffix multi-term OR query labels with _or by @vinaykakade in #10
• Add streaming to vector index builds by @ekechinwokah in #12
• Fix bad merge in #12 by @ekechinwokah in #13
• chore ci: add basic fmt, build & test on every PR by @Pranav2612000 in #11
• fts: leapfrog AND intersection over the skip table by @vinaykakade in #14
• Add lazy on-demand loads to vector reader by @ekechinwokah in #16
• Add streaming to FTS index builds by @ekechinwokah in #15
• fts: block-max-AND pruning + 2-term flat-merge specialization by @vinaykakade in #17
• test: stabilize commit::stats RSS readings under parallel exec by @ekechinwokah in #20
• improv ci: run tests only once ( during coverage check ) by @Pranav2612000 in #19
• fts: generalize flat-merge AND intersection to n >= 3 terms by @vinaykakade in #22
• Add rerank codecs for vectors by @ekechinwokah in #21
• test: tier-1 coverage push (+1.34 pp lines, 90.14% → 91.48%) by @vinaykakade in #23
• Add SIMD fast paths to vectors and remove Bf16 codec by @ekechinwokah in #24
• vector/distance: gate x86_64-only test helpers so make ci passes on aarch64 by @vinaykakade in #26
• examples: end-to-end demo across superfile, open-format, supertable by @muralikpbhat in #28
• WAL-driven updates + deletes by @vinaykakade in #27
• test: remove flaky whole-process RSS growth assertion by @vinaykakade in #31
• examples: name demo magic values as constants by @muralikpbhat in #30
• [1/n] improv superfile id_lookup: only read id rows from parquet by @Pranav2612000 in #33
• docs: add initial CONTRIBUTING guide and README Development section by @naman-modi in #35
• bench/supertable-update: correctly update using mutation counter across criterion calls by @vinaykakade in #36
• runtime: consolidate sync→async bridging into one bridge + runtime by @vinaykakade in #38
• Add object storage support by @ekechinwokah in #34
• Fix bug with retries on S3 connections by @ekechinwokah in #40
• docs: add AGENTS.md for AI agents (with CLAUDE.md symlink) by @ashishmishra26 in #37
• storage: add Azure Blob StorageProvider by @pratyushlokhande in #41
• improv fts: dedupe reader parsing and top-k plumbing by @naman-modi in #42
• Add SQL-native search via DataFusion by @ekechinwokah in #39
• Remove Criterion from benches by @ekechinwokah in #53
• Fix bug when running benchmarks on s3-fs emulator by @ekechinwokah in #51
• Polish up comments, constants, and license by @ekechinwokah in #45
• benches: explicit object-store backend selection + Azure (INFINO_BENCH_STORE) by @pratyushlokhande in #57
• Restore reader method on supertable by @ekechinwokah in #59
• supertable/query: FTS token_match + exact_match APIs and WHERE pushdown by @ekechinwokah in #58
• Extend SQL tests in benches by @ekechinwokah in #63
• storage: bring Azure resilience to S3 parity via shared retry path by @pratyushlokhande in #61
• ci: on-demand Azure supertable benchmark (in-region VM + sccache blob cache) by @pratyushlokhande in #62
• ci: scope bench VM to a fixed RG with exact per-run teardown by @pratyushlokhande in #65
• ci: run benchmark under the ci environment for branch-agnostic OIDC by @pratyushlokhande in #66
• ci: raise fd limit before the benchmark to avoid EMFILE on commit by @pratyushlokhande in #67
• benches: doc-count override is a plain integer, fix README example by @muralikpbhat in #68
• benches/ci: configurable bench selection + unified doc-count env by @pratyushlokhande in #69
• ci: drop inter-bench noise from the benchmark result summary by @pratyushlokhande in #72
• ci: run the bench group on every PR (informational) by @pratyushlokhande in #71
• ci(bench): compare PR benchmarks against main + track history by @pratyushlokhande in #73
• Public interface: Connection + Supertable catalog API, curated surface, Python bindings by @vinaykakade in #70
• feat superfile writer: add fn to create batch from another superfile by @Pranav2612000 in #60
• ci(bench): plot P90 RSS trend over last 10 main commits by @pratyushlokhande in #74
• chore: set up pre-commit hooks for cargo fmt & clippy checks by @Pranav2612000 in #79
• chore: compact config options added by @savannahar68 in #78
• feat superfile writer: ignore deleted documents when creating batch from superfile by @Pranav2612000 in #77
• chore: populate subsection_offsets.total_size for mutations by @savannahar68 in #81
• fix manifest commit: validates manifest_id before commiting manifest by @Pranav2612000 in #83
• feat superfile: create new superfile bytes by merging multiple readers by @Pranav2612000 in #80
• Refactor benches and extend public API by @ekechinwokah in #76
• feat: compaction policy v1 by @savannahar68 in #84
• improv superfile builder: return superfile stats when building from readers by @Pranav2612000 in #85
• fix benches: profile_segment calls renamed vector_hits_async by @naman-modi in #87
• feat supertable: merge superfileentries to new file's shardoutput by @Pranav2612000 in #89
• nit: remove action to build py package on macos13 from ci pipeline by @naman-modi in #93
• Feat/build merged shard by @Pranav2612000 in #94
• AGENTS.md: require imports at the top of the file by @vinaykakade in #95
• improv supertable commit: prevent redundant superfile writes during manifest CAS error by @Pranav2612000 in #96
• feat fts: -term negation for bm25_search by @naman-modi in #82
• feat: compact function to trigger compaction by @savannahar68 in #97
• supertable: drop redundant fp32 centroids from open_blob (-72% cold open) by @muralikpbhat in #101
• rename: 'segment' terminology to 'superfile' across the codebase by @vinaykakade in #98
• Revise search section to emphasize hybrid search by @ekechinwokah in #91
• [1/n] refactor supertable manifest: move manifest rebalancing to a separate function by @Pranav2612000 in #100
• fix chore pre-commit: modify clippy command to match ci by @Pranav2612000 in #105
• [2/n] feat manifest: support removing superfile entries from manifest by @Pranav2612000 in #102
• reader_cache: bg fill must re-check the reader after its grace sleep by @ekechinwokah in #106
• bench: prepare/benchmark split — run search on a pre-uploaded dataset by @pratyushlokhande in #92
• [3/n] feat supertable compaction: compact superfiles and update manifest for each job by @Pranav2612000 in #103
• fix: compaction select logic by @savannahar68 in #109
• chore: public api for compaction by @savannahar68 in #107
• bench: faithful cold vector measurement at supertable scale by @muralikpbhat in #88
• Fix a few issues discovered during testing by @ekechinwokah in #86
• Fix more testing issues: reduce aggregation latencies and clean up benchmark reporting by @ekechinwokah in #112
• feat: node bindings by @ashishmishra26 in #90
• Fix memory leak on supertable/query: break SessionContext cycle that leaked consumers on cold reopen by @ekechinwokah in #115
• Clean up repo for early OSS users by @ekechinwokah in #116
• parallelize corpus generation in benchmarks by @ekechinwokah in #117
• chore: Add regression test for O(n) commit latency with LocalFs storage by @savannahar68 in #118
• chore: node publish workflow by @ashishmishra26 in #120
• improv compaction tombstone: reduce tombstone network calls during compaction by @Pranav2612000 in #119
• ci: add manual workflow to publish the Python package to PyPI by @pratyushlokhande in #122
• chore: reduce encoding decoding per commit by @savannahar68 in #121
• ci: require SemVer for the Python release version by @pratyushlokhande in #125
• chore: update node publish workflow by @ashishmishra26 in #123
• docs: restructure README for onboarding — TOC, Install, agent quickstart, hybrid retrieval by @vinaykakade in #127
• Update package name by @ashishmishra26 in #130
• ci: speed up Python wheel builds; drop unsupported Windows target by @pratyushlokhande in #126
• docs: install the Python package from TestPyPI by @pratyushlokhande in #134
• Publish private npm package by @ashishmishra26 in #135
• chore: update publish workflow by @ashishmishra26 in #137
• a few fixes from 1B doc tests. by @ekechinwokah in #138
• test compaction: add compaction integration test by @Pranav2612000 in #129
• refactor manifest: expose manifest::load method for reading and loading manifest from storage by @Pranav2612000 in #139
• chore: [rebalance] minor optimisation of loading parts by @savannahar68 in #128
• test: raise line/function/region coverage above 90% by @vinaykakade in #157
• chore ci: fail ci if test coverage drops below 90 by @Pranav2612000 in #131
• ci: add automated PR reviewer by @pratyushlokhande in #136
• refactor manifest: allow manifest ops only through api by @Pranav2612000 in #159
• Revise README for clarity and feature updates by @ekechinwokah in #140
• feat: update & delete support in node by @ashishmishra26 in #161
• chore: allow one compaction at any given point in one process by @savannahar68 in #162
• python: show the README on the PyPI project page by @pratyushlokhande in #133
• ci: post bench results as a sticky PR comment by @pratyushlokhande in #169
• chore: retries in compaction flow duing occ errors by @savannahar68 in #165
• ci: split dataset bench into prepare + bench workflows by @pratyushlokhande in #168
• ci: fix and broaden the bench trend charts by @pratyushlokhande in #171
• ci: grant id-token write on dataset workflow callers by @pratyushlokhande in #175
• python: expose update, delete and compact by @pratyushlokhande in #167
• feat(fts): add Supertable::count — count-only matcher with O(1) df fast path by @vinaykakade in #160
• ci: post PR-Agent suggestions as inline comments by @pratyushlokhande in #177
• chore: [GC 1/n] add list with metadata function in storage by @savannahar68 in #174
• refactor manifest: rewrite manifest update fns as methods by @Pranav2612000 in #172
• python: expose cache and cold-fetch connect options by @pratyushlokhande in #178
• chore ci: fail ci if function coverage drops below 90% by @Pranav2612000 in #163
• refactor manifest: rename var for number of target_superfiles_per_part by @Pranav2612000 in #184
• catalog: drive table-free query_sql on a multi-thread runtime by @pratyushlokhande in #187
• Add Filtered Vector Search via Predicate Pushdown by @ekechinwokah in #173
• fix manifest rebalance: dont rebuild unchanged parts while removing entries by @Pranav2612000 in #186
• refactor manifest stats: store manifest list entry stats as arrayrefs by @Pranav2612000 in #190
• feat: add gc support by @savannahar68 in #185
• feat manifest stats: add methods for building & merging StatsAgg by @Pranav2612000 in #191
• chore ci: run pr review for all PRs by @Pranav2612000 in #203
• Add filtered vector search option in bindings and update docs by @ashishmishra26 in #200
• feat manifest stats: deprecate StatsTable by @Pranav2612000 in #202
• ci/bench: AI-summarized PR comment, only significant deltas by @pratyushlokhande in #204
• Update node package docs by @ashishmishra26 in #211
• examples: Python RAG examples for the infino package by @pratyushlokhande in #210
• chore: optimise fn wrapper on top of compact fn by @savannahar68 in #201
• python docs: document SQL search TVFs and hybrid search by @pratyushlokhande in #214
• ci: run the Python RAG examples end to end by @pratyushlokhande in #213
• ci: answer /ask replies on inline review comments by @pratyushlokhande in #216
• [1/n] refactor manifest: use Bloom struct in FtsSummaryAgg by @Pranav2612000 in #217
• [2/n] feat manifest stats: add methods for building, merging & using FtsSummaryAgg by @Pranav2612000 in #218
• [3/n] feat manifest stats: deprecate FtsSummary by @Pranav2612000 in #219
• docs/ci: document and verify uv install for the Python package by @pratyushlokhande in #221
• [4/n] feat manifests: add merge fn for vector aggs by @Pranav2612000 in #220
• ci/bench: skip reporting steps on cancel, keep teardown by @pratyushlokhande in #223
• Add examples in infino-node by @ashishmishra26 in #222
• feat manifest commit: rely on current stats for new stats calculation while building parts by @Pranav2612000 in #212
• improv: import hygiene; imports at top of file + grouped layout by @naman-modi in #224
• refactor manifests: rename rebalance fn to update by @Pranav2612000 in #227
• refactor manifests: rename ManifestListEntry to ManifestPartEntry by @Pranav2612000 in #225
• test: add locomo test by @ashishmishra26 in #226
• refactor manifests: rename Manifest->ManifestSnapshot & ManifestList->Manifest by @Pranav2612000 in #229
• docs: install infino from PyPI instead of TestPyPI by @pratyushlokhande in #232
• examples: add code-search, analytics, and inventory examples by @pratyushlokhande in #228
• docs: rename compact to optimize in Python/Node READMEs by @pratyushlokhande in #233
• feat manifest: add cache for manifest parts by @Pranav2612000 in #231
• python: add typed stubs, py.typed, and module metadata by @pratyushlokhande in #188
• feat: update node package to make it public by @ashishmishra26 in #241
• feat: compact(v1) sq8 vectors by @savannahar68 in #242
• feat: publish crate by @ashishmishra26 in #246
• feat manifest partition: use ingestion time with daily granularity as default partition by @Pranav2612000 in #245
• chore: make npm package public by @ashishmishra26 in #247

New Contributors

• @vinaykakade made their first contribution in #1
• @ekechinwokah made their first contribution in #7
• @muralikpbhat made their first contribution in #28
• @naman-modi made their first contribution in #35
• @ashishmishra26 made their first contribution in #37
• @pratyushlokhande made their first contribution in #41
• @savannahar68 made their first contribution in #78

Full Changelog: https://github.com/infino-ai/infino/commits/v0.1.0