Replace the old INSERT INTO t(rowid) VALUES('command') hack with a
proper hidden command column named after the table (FTS5 pattern):
INSERT INTO t(t) VALUES ('oversample=16')
The command column is the first hidden column (before distance and k)
to reserve ability for future table-valued function argument use.
Schema: CREATE TABLE x(rowid, <cols>, "<table>" hidden, distance hidden, k hidden)
For backwards compat, pre-v0.1.10 tables (detected via _info shadow
table version) skip the command column to avoid name conflicts with
user columns that may share the table's name. Verified with legacy
fixture DB generated by sqlite-vec v0.1.6.
Changes:
- Add hidden command column to sqlite3_declare_vtab for new tables
- Version-gate via _info shadow table for existing tables
- Validate at CREATE time that no column name matches table name
- Add rescore_handle_command() with oversample=N support
- rescore_knn() prefers runtime oversample_search over CREATE default
- Remove old rowid-based command dispatch
- Migrate all DiskANN/IVF/fuzz tests and benchmarks to new syntax
- Add legacy DB fixture (v0.1.6) and 9 backwards-compat tests
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Infrastructure improvements:
- Fix benchmarks-ann Makefile: type=baseline -> type=vec0-flat (baseline
was never a valid INDEX_REGISTRY key)
- Add DiskANN + text primary key test: insert, KNN, delete, KNN
- Add rescore + text primary key test: insert, KNN, delete, KNN
- Add WAL concurrency test: reader sees snapshot isolation while
writer has an open transaction, KNN works on reader's snapshot
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New benchmarks-ann/bench-delete/ directory measures KNN recall
degradation after random row deletion across index types (flat,
rescore, IVF, DiskANN). For each config and delete percentage:
builds index, measures baseline recall, copies DB, deletes random
rows, measures post-delete recall, VACUUMs and records size savings.
Includes Makefile targets, self-contained smoke test with synthetic
data, and results DB for analysis.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Extend benchmarks-ann/ with results database (SQLite with per-query detail
and continuous writes), dataset subfolder organization, --subset-size and
--warmup options. Supports systematic comparison across flat, rescore, IVF,
and DiskANN index types.
Add DiskANN graph-based index: builds a Vamana graph with configurable R
(max degree) and L (search list size, separate for insert/query), supports
int8 quantization with rescore, lazy reverse-edge replacement, pre-quantized
query optimization, and insert buffer reuse. Includes shadow table management,
delete support, KNN integration, compile flag (SQLITE_VEC_ENABLE_DISKANN),
release-demo workflow, fuzz targets, and tests. Fixes rescore int8
quantization bug.
Add inverted file (IVF) index type: partitions vectors into clusters via
k-means, quantizes to int8, and scans only the nearest nprobe partitions at
query time. Includes shadow table management, insert/delete, KNN integration,
compile flag (SQLITE_VEC_ENABLE_IVF), fuzz targets, and tests. Removes
superseded ivf-benchmarks/ directory.
Add approximate nearest neighbor infrastructure to vec0: shared distance
dispatch (vec0_distance_full), flat index type with parser, NEON-optimized
cosine/Hamming for float32/int8, amalgamation script, and benchmark suite
(benchmarks-ann/) with ground-truth generation and profiling tools. Remove
unused vec_npy_each/vec_static_blobs code, fix missing stdint.h include.
Add rescore index type: stores full-precision float vectors in a rowid-keyed
shadow table, quantizes to int8 for fast initial scan, then rescores top
candidates with original vectors. Includes config parser, shadow table
management, insert/delete support, KNN integration, compile flag
(SQLITE_VEC_ENABLE_RESCORE), fuzz targets, and tests.
Add approximate nearest neighbor infrastructure to vec0: shared distance
dispatch (vec0_distance_full), flat index type with parser, NEON-optimized
cosine/Hamming for float32/int8, amalgamation script, and benchmark suite
(benchmarks-ann/) with ground-truth generation and profiling tools. Remove
unused vec_npy_each/vec_static_blobs code, fix missing stdint.h include.
