tally 0.1.0 (2026-07-04)

First release. A from-scratch reimplementation of GNU wc(1), byte-identical
to coreutils 9.11 and faster on every workload measured.

Fused SIMD lines+words+bytes kernel (SSE2/AVX2/NEON) with decode-free
UTF-8 whitespace handling; default invocation 22x (ASCII) to 123x
(non-ASCII) over GNU wc on the reference x86-64 box.
Structurally validated -m character counting (68x on UTF-8 text).
Screened -L display-width scanner with a cached width table.
Full GNU surface: -l -w -c -m -L, --total, --files0-from, GNU
option parsing (permutation, abbreviations), locale and POSIXLY_CORRECT
semantics, shell-escape filename quoting, byte-exact diagnostics.
Installs as tally and ty.
Documented deviations only (man page DEVIATIONS section), including one
upstream bug fixed rather than replicated: GNU wc 9.11 miscounts words
when adjacent multibyte spaces split across read boundaries.

Provide feedback