feat: add Arinae algorithm#990

Merged

LoricAndre merged 80 commits intomasterfrom

Mar 1, 2026

Commits on Feb 21, 2026

feat: initial work on skim v3
LoricAndre
committed

Commits on Feb 23, 2026

Commits on Feb 25, 2026

Commits on Feb 26, 2026

feat: multiple optimizations
LoricAndre
committed
perf(skim_v3): use 2-row rolling buffer for score-only DP path
LoricAndre
committed
perf(skim_v3): add early termination when DP rows are all-zero
LoricAndre
committed
perf(skim_v3): add range_dp for fuzzy_match_range, avoiding full index vec
LoricAndre
committed
perf(skim_v3): remove redundant is_subsequence scan in exact mode
LoricAndre
committed
perf(skim_v3): avoid clone in traceback by using mem::take on thread-local buffer
LoricAndre
committed
perf(skim_v3): tighten typo-mode upper band bound in typo_vband_row
LoricAndre
committed
perf(skim_v3): use memchr SIMD for first-char search in prefilter and banding
LoricAndre
committed
revert(skim_v3): restore m upper bound in typo_vband_row
LoricAndre
committed
perf(skim_v3): add ASCII fast path to char::eq_ignore_case
LoricAndre
committed
perf(skim_v3): replace RefCell with UnsafeCell (TLCell) in thread-locals
LoricAndre
committed
fix(skim_v3): fix precompute_bonuses reserve logic
LoricAndre
committed
guard: return None for pat.len() > MAX_PAT_LEN in exact mode
LoricAndre
committed
perf: re-encode Dir::None=0 so CELL_ZERO is all-zero bytes
LoricAndre
committed
perf: 128-bit ASCII bitset for cheap_typo_prefilter tail scan
LoricAndre
committed
perf: early exit in count_tail_present_ordered when match is impossible
LoricAndre
committed
cleanup: remove unused constants SEPARATOR_MASK_LO/HI and FIRST_CHAR_BONUS_MULTIPLIER
LoricAndre
committed
refactor: replace unsafe transmute in Cell::dir() and compute_cell with safe match
LoricAndre
committed
perf: Atom::is_sep() trait method avoids u8→char→u32 in separator check
LoricAndre
committed
refactor: precompute_bonuses rewritten as safe iterator chain
LoricAndre
committed
refactor: extract match_slices_range; simplify run_range
LoricAndre
committed
mem: SWMatrix::resize shrinks when buffer is 4× over-allocated
LoricAndre
committed
Revert "mem: SWMatrix::resize shrinks when buffer is 4× over-allocated"
LoricAndre
committed
Revert "refactor: replace unsafe transmute in Cell::dir() and compute_cell with safe match"
LoricAndre
committed
Revert "perf: Atom::is_sep() trait method avoids u8→char→u32 in separator check"
LoricAndre
committed
Revert "perf: early exit in count_tail_present_ordered when match is impossible"
LoricAndre
committed
Revert "perf: 128-bit ASCII bitset for cheap_typo_prefilter tail scan"
LoricAndre
committed
Revert "refactor: extract match_slices_range; simplify run_range"
LoricAndre
committed
Revert "perf(skim_v3): replace RefCell with UnsafeCell (TLCell) in thread-locals"
LoricAndre
committed
Revert "perf(skim_v3): add ASCII fast path to char::eq_ignore_case"
LoricAndre
committed
Revert "revert(skim_v3): restore m upper bound in typo_vband_row"
LoricAndre
committed
Revert "perf(skim_v3): tighten typo-mode upper band bound in typo_vband_row"
LoricAndre
committed
Revert "perf(skim_v3): avoid clone in traceback by using mem::take on thread-local buffer"
LoricAndre
committed
Revert "perf(skim_v3): add early termination when DP rows are all-zero"
LoricAndre
committed
Revert "perf(skim_v3): use 2-row rolling buffer for score-only DP path"
LoricAndre
committed
fix: reverse only order of frizbee indices
LoricAndre
committed
chore: rename & refactor into multiple files
LoricAndre
committed
Merge branch 'master' into skim-v3
LoricAndre
authored
chore: optimizations to the main flow
LoricAndre
committed
fix: correct banding in non-typo path
LoricAndre
committed
chore: generate completions & manpage
Skim bot
committed
docs: add algorithms section to the README [skip ci]
LoricAndre
committed
fix(ari): correctly bound vband low
LoricAndre
committed

Commits on Feb 27, 2026

Commits on Mar 1, 2026

chore: review
LoricAndre
committed