Download Latest Version 0.90.0 source code.zip (8.9 MB) Google Add to Preferred Sources
Home / 0.88.0
Name Modified Size InfoDownloads / Week
Parent folder
vx-x86_64-apple-darwin.tar.gz 2026-10-08 39.2 MB
0
vx-aarch64-apple-darwin.tar.gz 2026-10-08 36.9 MB
0
vx-aarch64-unknown-linux-gnu.tar.gz 2026-10-08 35.1 MB
0
vx-x86_64-unknown-linux-gnu.tar.gz 2026-10-08 37.4 MB
0
0.88.0 source code.tar.gz 2026-10-08 6.2 MB
0
0.88.0 source code.zip 2026-10-08 8.6 MB
0
README.md 2026-10-08 5.6 kB
0
Totals: 7 Items   163.3 MB 0

Important breaks

None

Changes

⚠️ Breaks

  • Speed up scalar construction, stats reads and aggregate partial merging (#10326) @robert3005
  • break(arrow): add export options and allow skipping view buffer compaction (#10220) @lorenzhs
  • Blocked BitPacked: store block offsets as a child (#10004) @mhk197
  • Delta validity is a top level child (#9970) @robert3005
  • Move users of map_each_with_validity and map_each_in_place to lane kernels and remove those methods (#10139) @robert3005

🚧 Deprecation

  • Remove usages of null_as_false (#10124) @myrrc

✨ Features

  • Add support for reading vortex files through native python readers (#10097) @robert3005
  • duckdb: don't execute validity in exporters (#10376) @myrrc
  • Blocked BitPacked: encode blocked bitwidths (#10203) @mhk197
  • AllFalse early exit (#10303) @myrrc
  • Scan protocol traits (#10047) @joseph-isaacs
  • Add combine_chunks option to Python Arrow export (#10279) @robert3005
  • feat(python): support integer column indices in read_url projection (#10287) @jackylee-ch

🚀 Performance

  • Bool take doesn't fill_null validity (#10383) @myrrc
  • perf(array): expand FixedSizeList filter masks directly into a bitmap (#10274) @robert3005
  • Make opening a Vortex file with a cached footer cheaper (#10358) @joseph-isaacs
  • perf: classify canonical arrays by concrete vtable (#10238) @joseph-isaacs
  • Add OnPair take kernel that shares the pair dictionary (#10363) @joseph-isaacs
  • perf(arrow): preserve source nullability during export (#10365) @joseph-isaacs
  • Read integer validity a word per chunk, and fill nulls for runs (#10342) @joseph-isaacs
  • Read VarBin UTF-8 RowFn inputs through their offsets (#10340) @connortsui20
  • Execute FSST iteratively (#10209) @joseph-isaacs
  • perf(array): avoid errors when comparison kernels decline arithmetic (#10301) @connortsui20
  • Faster ScalarFnArray creation, lazy ArrayStats (#10329) @myrrc
  • perf(buffer): use fearless_simd for popcount (#10327) @joseph-isaacs
  • Initialise ArrayParts directly with provided slots (#10072) @robert3005
  • Execute OnPair iteratively (#10210) @joseph-isaacs
  • perf(compressor): detect constant primitives before generating stats (#10043) @robert3005
  • perf: vectorize small u8 table take with AVX2 (#9572) @joseph-isaacs
  • perf: vectorize small u8 table take with NEON (#9571) @joseph-isaacs
  • Optimise min/max computation for VarBinView arrays (#9740) @robert3005
  • Execute DecimalByteParts iteratively (#10207) @joseph-isaacs
  • Execute RunEnd iteratively (#10206) @joseph-isaacs
  • perf(zstd): reuse one decompression context per thread (#10270) @joseph-isaacs
  • feat: add between kernel for DecimalByteParts (#10152) @joseph-isaacs
  • python: release the GIL only around expensive work (#10280) @robert3005
  • Convert trivial filters into slices during reduction, MaskValues::last uses BitBuffer::last_set_index (#9831) @robert3005
  • perf(array): forward probe_scalar through pass-through encodings (#9905) @joseph-isaacs
  • Perform validation and null view replacement in two passes (#10148) @robert3005

🐛 Bug Fixes

13 changes * fix(python): translate Polars time literals (#10186) @danking * fix(buffer): require both halves of a Zip to be TrustedLen (#10293) @jianhe25 * fix(buffer): validate offsets in Buffer::into_arrow_offset_buffer (#10292) @jianhe25 * fix(buffer): don't create &mut [u64] over uninitialized memory in collect_words_in (#10294) @jianhe25 * Trim child data on Arrow slice imports (#10379) @palaska * duckdb: i16 Decimal on duckdb side may fit in Vortex's i8 decimal buffer (#10155) @myrrc * Keep the file sum absent when a chunk sum overflows (#10328) @connortsui20 * Ignore empty constant chunks in aggregate states (#10254) @connortsui20 * fix(parquet-variant): decline all_non_distinct when only one side is shredded (#10275) @jackylee-ch * fix(python): preserve nullability when converting a chunked array with nulls (#10283) @jackylee-ch * fix(buffer): enable and detect bmi1 for _tzcnt_u64 in select_in_word (#10295) @jianhe25 * fix(array): preserve result nullability when simplifying and/or/zip (#10286) @jackylee-ch * fix(datafusion): preserve list element field metadata in physical schema (#10285) @jackylee-ch

🧰 Maintenance

14 changes * Skip validation when accessing storage scalar of extension scalar (#9875) @robert3005 * Validate TPC-DS results against Vortex and Parquet (#9863) @joseph-isaacs * Test IO file sizes in Rust and array sizes in Python doctests (#9792) @robert3005 * chore(array): use fearless_simd for AVX2 1- and 2-byte filter compress (#10324) @joseph-isaacs * Add sum oracle to the fuzzer and reject samples that have order dependent overflow behaviour (#9801) @robert3005 * Do not build a Sequence array for slices that are not sequences (#10338) @robert3005 * perf(buffer): portable fearless_simd kernels for bitmap rank-select (#10319) @joseph-isaacs * fix: reconcile asyncband references in Cargo.lock (#10306) @joseph-isaacs * Fix deprecations flagged by cargo build on new rust version (#10042) @robert3005 * Add list_contains membership benchmarks (#10060) @robert3005 * Clean up `vortex_ensure_eq!` (#10237) @connortsui20 * Lock file maintenance (#10290) @[renovate[bot]](https://github.com/apps/renovate) * Split SQL PR benchmarks into bench-sql and bench-sql-extended (#10213) @joseph-isaacs * Update bit-vec to 0.10.1 (#10288) @jianhe25
Source: README.md, updated 2026-10-08