fix(interpreter): ${#v} counts bytes under the C locale

[chaliy]

Requested by Михайло · project thread

What changed

• ${#v} counts bytes when the shell's locale is C or POSIX, read from LC_ALL, then LC_CTYPE, then LANG (first non-empty), as bash does. local LC_ALL=... scopes it like any variable.
• ${#v} inside a default operand (${x:-${#s}}) counted bytes in every locale; it now counts characters like the top-level form.

Why

Bash-oracle gap unicode-and-bytes. Agent scripts set LC_ALL=C for fast sort/grep and then size strings; bashkit reported characters where bash reports bytes.

Before / After

$ LC_ALL=C; s=ação; echo ${#s}
before: 4      after: 6      (bash 5.2: 6)
$ s=ação; echo ${x:-${#s}}
before: 6      after: 4      (bash 5.2: 4)

New spec c-locale-length.test.sh (5 cases, checked against bash 5.2 and by bash_comparison_tests).

Risk

Low. With no locale variable set bashkit stays UTF-8 (bash would fall back to C). Slices, case changes and globs stay character-based under C because shell words cannot hold a split UTF-8 sequence; documented in limitations ("Locale").

Checklist

• fmt, clippy -D warnings
• integration and lib tests, bash_comparison_tests
• knowledge (log.md, limitations)

---

Generated by Claude Code

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	bashkit	e68f9f4	Commit Preview URL Branch Preview URL	Oct 08 2026, 03:00 AM