Skip to content

CHANGELOG.md

Latest commit

 

History

History
58 lines (45 loc) · 2.03 KB

CHANGELOG.md

File metadata and controls

58 lines (45 loc) · 2.03 KB

Changelog

git

  • docs: small clarifications

2.3.1

  • bugfix: Fix bug on 32-bit platforms where 1<<32 overflows.

2.3.0

  • feat: add support for open and closed syncmers.
  • code: simplify/cleanup dedup intrinsics
  • ci: skip some tests in debug mode
  • misc: cargo update to get ensure_simd

2.2.0

  • Feat: add run_with_buf that reuses internal buffers.
  • Perf: more inline(always).
  • Perf: avoid u32x8::splat on NEON.
  • Doc: clarify target-cpu=native in readme.

2.1.0

  • Breaking:: bump packed-seq to 4.1 with PackedNSeq.
  • Add scalar feature that forwards to packed-seq, to disable warnings about lack of SIMD features when native compilation is not enabled.

2.0.1 (yanked)

  • Drop accidental #[feature(type-changing-struct-update)]

2.0.0 (yanked)

  • Breaking: Migrate to packed-seq 4.0 with PaddedIt.
  • Breaking: Improve simd versions for short sequences by reusing allocated buffers.
  • Feature: Use seq-hash crate to cleanly support multiple kmer hashers; add *_with_hasher function variants.
  • Feature: Add support for _skip_ambiguous_windows variants that omit minimizers for windows containing non-ACTG (such as NYR).
  • Move simd-minimizers crate from subdir into the repo root.
  • Cleanups (use Rust 2024 edition; drop some public-but-unused functions).
  • Improve simd versions for short sequences by preventing over-allocating the output vector.

1.4

  • Make NtHash and MulHash seedable.

1.3

  • Update to packed-seq 3.2.1 for u128 kmer value support.
  • Add iter_{canonical}_minimizer_values_u128 to iterate over u128 kmer values.

1.2

  • Fix #10: Add iter_{canonical}_minimizer_values to convert positions into u64 kmer values.
  • Update to packed-seq 3.0.
  • Fix to properly initialize arrays when collecting super-k-mers.
  • Update packed-seq to support non-byte offsets.

1.1

  • Update packed-seq to 2.0, which uses tuples of (simd iterator, padding), instead of a separate scalar iterator over the tail.

1.0

  • Initial release.