max / makenotwork

tagtree: count max_length in characters, not bytes TagConfig, TagError::TooLong, and the crate docs all describe max_length in characters; validate_with measured bytes. A multi-byte tag under the limit was reported as TooLong rather than InvalidChar. Byte length bounds character count from above, so the cheap check still gates the O(n) one and the ASCII path is unchanged. Non-ASCII tags were rejected before and are rejected now; only the reported error moves.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: Max Johnson <me@maxj.phd> · 2026-07-20 17:57 UTC

Signed with PGP, not checked

Commit: 36254a8e1e018e4ab14d61cad9f672a0ed10270b

Parent: 2fc5764

2 files changed, +33 insertions, -1 deletion

M shared/tagtree/CHANGELOG.md +9

			@@ -2,6 +2,15 @@
2	2
3	3		Notable changes to tagtree. Versions follow semver.
4	4
	5	+	## Unreleased
	6	+
	7	+	### Fixed
	8	+
	9	+	- `max_length` now counts characters, matching what `TagConfig`, `TagError::TooLong`,
	10	+	and the crate docs have always said. It measured bytes, so a multi-byte tag under
	11	+	the limit was reported as `TooLong` instead of `InvalidChar`. Only the reported
	12	+	error changes: non-ASCII tags were rejected before and are rejected now.
	13	+
5	14		## 0.4.0 (2026-07-16)
6	15
7	16		### Breaking

M shared/tagtree/src/lib.rs +24 -1

			@@ -141,7 +141,10 @@
141	141		if tag.is_empty() {
142	142		return Err(TagError::Empty);
143	143		}
144		-	if tag.len() > config.max_length {
	144	+	// `max_length` is documented in characters. Byte length is an exact upper
	145	+	// bound on character count, so the cheap check gates the O(n) one: only a
	146	+	// tag whose bytes overflow can possibly overflow in characters.
	147	+	if tag.len() > config.max_length && tag.chars().count() > config.max_length {
145	148		return Err(TagError::TooLong { max: config.max_length });
146	149		}
147	150		if tag.starts_with(SEPARATOR) \|\| tag.ends_with(SEPARATOR) {
			@@ -1026,6 +1029,26 @@
1026	1029		assert!(validate_with("abcdefghijk", &cfg).is_err()); // 11 chars
1027	1030		}
1028	1031
	1032	+	#[test]
	1033	+	fn max_length_counts_characters_not_bytes() {
	1034	+	let cfg = TagConfig {
	1035	+	max_depth: 10,
	1036	+	max_length: 5,
	1037	+	semantic_depth: 0,
	1038	+	};
	1039	+	// 3 chars, 9 bytes. Over the limit in bytes, under it in characters,
	1040	+	// so the charset check gets to report the real problem.
	1041	+	assert_eq!(
	1042	+	validate_with("日本語", &cfg),
	1043	+	Err(TagError::InvalidChar('日'))
	1044	+	);
	1045	+	// Still TooLong when the character count genuinely overflows.
	1046	+	assert_eq!(
	1047	+	validate_with("日本語日本語", &cfg),
	1048	+	Err(TagError::TooLong { max: 5 })
	1049	+	);
	1050	+	}
	1051	+
1029	1052		#[test]
1030	1053		fn config_enforces_semantic_depth() {
1031	1054		// semantic_depth=1 requires at least 2 segments