v0.23.1-beta.5

What's Changed

New Features 🎉

• feat: expose specifying scanner filters via datafusion by @westonpace in #3458

Full Changelog: v0.23.1-beta.4...v0.23.1-beta.5

v0.23.1-beta.4

What's Changed

Bug Fixes 🐛

• fix: allocate much memory for residual vectors than needed by @BubbleCal in #3446
• fix: flat KNN column stats order doesn't match schema by @BubbleCal in #3451

Performance Improvements 🚀

• perf: make miniblock decoding cheaper by @westonpace in #3438

Full Changelog: v0.23.1-beta.3...v0.23.1-beta.4

v0.23.1-beta.3

What's Changed

Bug Fixes 🐛

• fix: don't eagerly materialize fields that the user hasn't asked for by @westonpace in #3442

Full Changelog: v0.23.1-beta.2...v0.23.1-beta.3

v0.23.1-beta.2

What's Changed

Bug Fixes 🐛

• fix: implement with_new_children for FTS by @BubbleCal in #3441

Full Changelog: v0.23.1-beta.1...v0.23.1-beta.2

v0.23.1-beta.1

What's Changed

Bug Fixes 🐛

• fix: remove extraneous padding in plain encoder by @wkalt in #3434

New Contributors

• @wkalt made their first contribution in #3434

Full Changelog: v0.23.0...v0.23.1-beta.1

v0.23.0

What's Changed

Breaking Changes 🛠

• fix!: delta index fragment bitmaps contained previous index coverage by @wjones127 in #3377

New Features 🎉

• feat: enable all datafusion functions by @westonpace in #3381
• feat: support float16/float64 for multivector by @BubbleCal in #3387
• feat: add drop_index by @westonpace in #3382
• feat: add an all null column as a metadata-only operation by @albertlockett in #3391
• feat: finish up variable-length encodings in the full-zip path by @westonpace in #3344
• feat: add testing of string/binary to 2.1 full-zip encoding and fix bugs by @westonpace in #3418
• feat: auto-migrate old index metadata by @wjones127 in #3428
• feat: allow replacement of entire datafile when the schema lines up correctly by @chebbyChefNEQ in #3408

Bug Fixes 🐛

• fix: json schema serializes field metadata by @albertlockett in #3379
• fix: flat FTS would return all unindexed rows by @BubbleCal in #3386
• fix: updating schema/field metadata now retains fragments by @albertlockett in #3384
• fix: full text search index may be corrupted after remapping by @BubbleCal in #3388
• fix: handle the possibility that serialize_expressions returns a memoryview by @westonpace in #3396
• fix: merge_insert with subcols sometimes outputs unexpected nulls by @wjones127 in #3407
• fix: ensure that 'block_size' parameter is properly propagated in the ObjectStore by @vjc578db in #3403
• fix(rust): loosen bytemuck pin by @wjones127 in #3413
• fix: don't compare metadata in merge insert to detect if partial schema by @westonpace in #3412
• fix: support fp16 type in SQ by @chebbyChefNEQ in #3417
• fix: move IO tasks off of CPU runtime in merge_insert by @wjones127 in #3420
• fix: filter out null values when sampling for index training by @wjones127 in #3404
• fix: handle null vectors in flat search by @wjones127 in #3422
• fix: avoid divide-by-zero when training an index with a large dimension by @westonpace in #3426
• fix: bump openssl for CVE by @chebbyChefNEQ in #3431

Performance Improvements 🚀

• perf: avoid re-alloc on assigning PQ by @BubbleCal in #3399
• perf: skip shuffling if there is only 1 partition by @BubbleCal in #3405

New Contributors

• @vjc578db made their first contribution in #3403

Full Changelog: v0.22.0...v0.23.0

v0.23.0-beta.5

What's Changed

New Features 🎉

• feat: allow replacement of entire datafile when the schema lines up correctly by @chebbyChefNEQ in #3408

Bug Fixes 🐛

• fix: bump openssl for CVE by @chebbyChefNEQ in #3431

Full Changelog: v0.23.0-beta.4...v0.23.0-beta.5

v0.23.0-beta.4

What's Changed

New Features 🎉

• feat: auto-migrate old index metadata by @wjones127 in #3428

Bug Fixes 🐛

• fix: avoid divide-by-zero when training an index with a large dimension by @westonpace in #3426

Full Changelog: v0.23.0-beta.3...v0.23.0-beta.4

v0.23.0-beta.3

What's Changed

New Features 🎉

• feat: finish up variable-length encodings in the full-zip path by @westonpace in #3344
• feat: add testing of string/binary to 2.1 full-zip encoding and fix bugs by @westonpace in #3418

Bug Fixes 🐛

• fix: don't compare metadata in merge insert to detect if partial schema by @westonpace in #3412
• fix: support fp16 type in SQ by @chebbyChefNEQ in #3417
• fix: move IO tasks off of CPU runtime in merge_insert by @wjones127 in #3420
• fix: filter out null values when sampling for index training by @wjones127 in #3404
• fix: handle null vectors in flat search by @wjones127 in #3422

Full Changelog: v0.23.0-beta.2...v0.23.0-beta.3

v0.23.0-beta.2

What's Changed

Bug Fixes 🐛

• fix: ensure that 'block_size' parameter is properly propagated in the ObjectStore by @vjc578db in #3403
• fix(rust): loosen bytemuck pin by @wjones127 in #3413

Performance Improvements 🚀

• perf: skip shuffling if there is only 1 partition by @BubbleCal in #3405

New Contributors

• @vjc578db made their first contribution in #3403

Full Changelog: v0.23.0-beta.1...v0.23.0-beta.2