[derek73]

Part of #762. M2 (20 entries) is left for its own batch; lines that rest on an M2 decision point there with Elsewhere:.

What changes

• Standing blocks for decisions/M1 (delimited maiden names), M3 (the marker-led clause) and M4 (a maiden marker makes the lone name word the family). Each block has one present-tense line per decision in force, each with a pointer to its log entry, plus Not standing:, Elsewhere: and Open: lines and a ## Log heading. All three are taken off tests/v2/decisions_without_standing.json.
• Edit-kind-1 notes on log heads that later decisions reversed in part:
  • M1 2026-07-03: the opt-in boundary, narrowed by M3's Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 entry.
  • M1 Delimited maiden keeps its marker where the bare form drops it ((née Jones) vs née Jones) #329: the "multi-word clause" condition and the "clause-size guard", corrected by parse("Maria Kowalska z domu Nowak") cannot reach the Polish maiden marker — markers do not compose the way suffixes do #434.
  • M3 Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 accepted-consequence: the bare née V contrast, changed by M2's Rethink the maiden-marker rule (M2): a marker stands after the current name and takes words up to the trailing suffix run #601.
  • M3 Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 implemented-list: _extract.py alone, extended by A nickname loses its comma: Jane (Janey, Jo) Smith gives nickname Janey Jo (1.4.0: Janey, Jo) #653.
  • M3 Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 durable-finding: the comma witness, moved by A nickname loses its comma: Jane (Janey, Jo) Smith gives nickname Janey Jo (1.4.0: Janey, Jo) #653.
  • M4 parse("Smith née Jones") reports no family name — a maiden clause empties a one-word name, where 1.4.0 kept it #445 precedence and corpus-property entries: the 'Smitty' Jones Jr. née Smith witness, changed by Rethink the maiden-marker rule (M2): a marker stands after the current name and takes words up to the trailing suffix run #601.
• rules/M3's Accepted bullet was stale. Its contrast was Jane Smith (née V) against the bare Jane Smith née V, but since Rethink the maiden-marker rule (M2): a marker stands after the current name and takes words up to the trailing suffix run #601 (decisions/M2) the bare spelling reads maiden 'V' too. The contrast is now Jane Smith (née PhD) → maiden 'PhD' against Jane Smith née PhD → suffix 'PhD'. The (née V) row moves to the main fence, where its reading still holds. corpus_rules.jsonl and snapshot.jsonl gain the two new names and nothing else moves.
• Checker (tests/v2/design_docs.py).
  • M3 and M4 each have several entries whose heads are the same words (2026-08-26 #335, 2026-08-27 #445) beside one head with that date and issue that goes on longer (#335 (M3, PR on …)). The resolver refused to resolve those twin heads by their text key; it now does.
  • Not standing: pointers that name more than one entry now fail as ambiguous, as Standing pointers already did.
  • New rows in test_design_docs.py cover both directions. The README and the standing_findings docstring say what is checked.

Verification

• Every cited example was parsed on the tree, under all three name orders where the claim depends on order, and with Policy(maiden_delimiters=frozenset({("(", ")")})) where it matters.
• Every Standing and Not-standing pointer resolves to exactly one live entry.
• Full suite on 3.11 and on 3.15: 22742 passed. ruff and mypy are clean.

Review findings (two design-docs-reviewer rounds, both fixed)

Round 1:

1. The wider twin resolution let a typo'd key on a Not standing: line pass silently, because that branch never checked for ambiguity. It now fails, and has test rows.
2. rules/M3's Accepted fence still showed (née V) as a bracketed-versus-bare contrast. Fixed as described above.
3. M1's Not-standing line pointed at the standing 2026-08-26 #335 (M3) entry. It now names Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 in prose.
4. Order-dependent fields were stated as if they held under every order (abd née Jones, j nee smith, Jane Smith (née Jr.), Jane Smith née PhD). They are now scoped to the default order.

Round 2, on the fix commit:

1. Three more order-dependent examples needed the same scoping: Jane (née Jones) Smith, Rosalind Roz Smith, 'Smitty' Jones Jr. née Smith. Scoped.
2. The standing_findings docstring and the README's ## Standing bullet did not say what a Not-standing pointer is checked for. Both now do.
3. M3's Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 accepted-consequence head now says the Accepted witness is (née PhD).
4. The reviewer confirmed the new Not-standing check fires on no existing file, ran each new row against both earlier checkers, and found no unmarked reversal in other files.

Stale, found and left

• nameparser/_policy.py's apply_patch docstring says the maiden-wins canonicalization was "decided 2026-07-19". decisions/M1's entry and commit cc7063f both say 2026-07-18, "ruled the same day". I didn't resolve which date is right.
• Several M1, M3 and M4 log entries reason from the differential gates that Replace the differential ledgers with a committed snapshot of corpus readings #664 retired (M3's condition entry cites the 2.1.0 gate going red; M4's ledger-markers entry; M1's two Parenthesized (née Jones) parses as a nickname where the bare née Jones gives maiden #335 gate-visibility entries). The decisions themselves still hold, so the entries are left as written, with Not standing: lines where an entry was only about the gates.
• M4's corpus-property entry says the maiden-clause property was falsified for "fourteen names" on 2026-08-27. Since Rethink the maiden-marker rule (M2): a marker stands after the current name and takes words up to the trailing suffix run #601 that count probably differs; the head is marked, not re-measured.
• M1's undated bullet "M1 and M2 are disjoint by construction" is no entry, so it has no Standing line.
• No H-section edits were needed (the H1–H6 batch is running in parallel).

Not merging; Derek decides.

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.05%. Comparing base (2b5e96b) to head (9f03c9a).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #814   +/-   ##
=======================================
  Coverage   99.05%   99.05%           
=======================================
  Files          47       47           
  Lines        4659     4659           
=======================================
  Hits         4615     4615           
  Misses         44       44           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.