chore(release): open the 2.3 cycle, and record the roster AGENTS.md missed

Release checklist steps 8 and 9, run after 2.2.0 went to PyPI.

- expected_since_2.2.0.toml, opened empty. DEFAULT_BASELINE and the
  ledger are coupled: _allowlist_for treats a missing file as a hard
  error, so the baseline cannot advance without it.
- compare.py DEFAULT_BASELINE 2.1.0 -> 2.2.0.
- VERSION (2,3,0) with PRE_RELEASE 'dev', so an install from master
  normalizes to 2.3.0.dev0 -- above 2.2.0, below 2.3.0, and never able
  to masquerade as the release it precedes. The docstring's own example
  moved with it. CITATION.cff stays at 2.2.0 by design: it names the
  last citable RELEASED artifact and lags for the whole cycle.
- "* 2.3.0 - Unreleased" opened at the top of the release log.

AGENTS.md step 8 named four rosters to enrol a new ledger in and I hit a
fifth, _CROSS_RULE_WINNERS, as a red run after the other registrations
were already in. Measured which are mandatory rather than guessing:
THREE are keyed by filename and asserted by EQUALITY, so every ledger
must appear even while empty (_SPAN_BEARING_RULES, _CORPUS_CLAIMS,
_CROSS_RULE_WINNERS), and TWO are keyed by rule content and genuinely
conditional (_HONORIFIC_SOURCES, _LATIN_ALTERNATION_SOURCES). Step 8
said "_CORPUS_CLAIMS always, and the others where they apply", which is
wrong for two of the five and had _HONORIFIC_SOURCES filed with the
mandatory ones. Rewritten under Required/Conditional headings, with
_CROSS_RULE_WINNERS added and #452's reason for its equality check.

Also records, at _CORPUS_CLAIMS, what a pure corpus RENAME looks like
there -- names unchanged, roles unchanged, digest moved -- since that
came up renaming de Mesnil Juan to Jean and the guard's message explains
a growing count but not a moving digest.

Verified: 6198 tests, mypy and ruff clean, docs clean. The gate against
the newly published 2.2.0 reports intentional diffs 0, unexplained 0 --
master is byte-identical to what shipped, which is the correct reading
on day one. 1.4.0 and 2.1.0 exit 0 at 226 and 106.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: derek73

AGENTS.md

@@ -139,20 +139,34 @@ uv run sphinx-build -b html docs dist/docs
 #    so a copy spelled some other way is not undeclared but unseen. What
 #    covers those is _CORPUS_CLAIMS, which records what every rule claims
 #    -- its regex's corpus reach, its roles, and which names -- and so
-#    needs no notion of how a copy is spelled. Enrol the new ledger in
-#    _CORPUS_CLAIMS always, and in the others where they apply:
+#    needs no notion of how a copy is spelled.
+#    THREE rosters are keyed by FILENAME and checked by EQUALITY, so a new
+#    ledger must be enrolled in every one of them on the day it lands, even
+#    empty -- each fails loudly and names itself, but that is three separate
+#    red runs if you add them one at a time. Two more are keyed by rule
+#    CONTENT and apply only where such a rule exists.
+#    Required, by filename:
 #      - _SPAN_BEARING_RULES: add the filename, mapped to the set of issue
 #        tags whose rules carry a script-span class (empty set if none).
+#      - _CORPUS_CLAIMS: a ledger with no entry hard-fails. Add the
+#        filename mapped to {} while the ledger is empty, then one _Claim
+#        per rule as rules land. The test prints the values to record. A
+#        _Claim's digest moves whenever the corpus text does, so a pure
+#        RENAME of a corpus name lands here as names-unchanged,
+#        roles-unchanged, digest-moved; a count that GREW means the rule
+#        absorbed something.
+#      - _CROSS_RULE_WINNERS: add the filename mapped to {} while the
+#        ledger has no contested name. Equality since #452, which found
+#        that a ledger with no rows had been indistinguishable from one
+#        needing none -- the two 2.x ledgers whose shapes #452 moved
+#        between rules had no section at all.
+#    Conditional, by rule content:
 #      - _HONORIFIC_SOURCES: if the ledger has a CJK honorific rule, add a
 #        substring of its issue (keyed that way, not by tag) mapped to the
 #        constant it copies -- but only if no existing key already matches
 #        that issue. A retroactive ledger can repeat an older one's rule
 #        verbatim (fix(#271/#272/#298) is in both today), and every rule
 #        must match exactly one key.
-#      - _CORPUS_CLAIMS: REQUIRED, not conditional -- a ledger with no
-#        entry hard-fails. Add the filename mapped to {} while the ledger
-#        is empty, then one _Claim per rule as rules land. The test prints
-#        the values to record.
 #      - _LATIN_ALTERNATION_SOURCES: same, for a rule copying a Latin
 #        vocabulary (maiden markers, ambiguous acronyms). An alternation
 #        matching no key fails as undeclared -- add it, or record it in

docs/release_log.rst

@@ -1,5 +1,7 @@
 Release Log
 ===========
+* 2.3.0 - Unreleased
+
 * 2.2.0 - August 31, 2026
 
     nameparser 2.2 is a rename plus about thirty parsing fixes.

nameparser/_version.py

@@ -7,13 +7,13 @@
 #: cannot carry the dev marker, so `VERSION >= (2, 2, 0)` is already
 #: true here while 2.2.0 is unreleased. Compare `__version__` instead
 #: where that distinction matters.
-VERSION = (2, 2, 0)
+VERSION = (2, 3, 0)
 #: PEP 440 pre-release/dev segment appended to the numeric version, or
 #: "" for a final release. Joined WITHOUT a dot ("2.0.0rc1", not
 #: "2.0.0.rc1"); setuptools reads __version__ as the package version.
 #:
 #: "dev" through the cycle, cleared at release. It normalizes to
-#: 2.2.0.dev0, which sorts above 2.1.0 and BELOW 2.2.0, so an install
+#: 2.3.0.dev0, which sorts above 2.2.0 and BELOW 2.3.0, so an install
 #: from master can never masquerade as the release it precedes.
-PRE_RELEASE = ""
+PRE_RELEASE = "dev"
 __version__ = ".".join(map(str, VERSION)) + PRE_RELEASE

tests/v2/test_ledger_guards.py

@@ -352,7 +352,8 @@ def test_script_ranges_membership_is_decided() -> None:
         "fix(#271/#272/#298)",              # the canonical class
         "fix(#298)",                        # the 间隔号 lookahead
     }),
-    "expected_since_2.1.0.toml": frozenset(),   # open cycle, no rules yet
+    "expected_since_2.1.0.toml": frozenset(),   # 2.2 cycle: no span-bearing rule
+    "expected_since_2.2.0.toml": frozenset(),   # open cycle, no rules yet
 }
 
 #: The leading `fix(...)`/`feat(...)` tag of a rule's `issue`, which is
@@ -1776,6 +1777,7 @@ def _claim(rule: dict) -> _Claim:
         "fix(#371) a suffix never begins a name: the Ph./D. merge declines at the head":
             _Claim(4, ('family', 'given', 'middle', 'suffix', 'title'), "1425d85a2d86"),
     },
+    "expected_since_2.2.0.toml": {},   # open cycle, no rules yet
     "expected_since_2.1.0.toml": {
         "fix(#371) a suffix never begins a name: the Ph./D. merge declines at the head":
             _Claim(4, ('family', 'given', 'middle', 'suffix', 'title'), "1425d85a2d86"),
@@ -1973,6 +1975,7 @@ def test_every_rule_claims_the_recorded_share_of_the_corpus() -> None:
 #: Re-measure rather than adjust them if a parser change moves one:
 #: a diff shape that shifted is a finding, not a number to update.
 _CROSS_RULE_WINNERS: dict[str, dict[tuple[str, tuple[str, ...]], str]] = {
+    "expected_since_2.2.0.toml": {},   # open cycle, no rules and so no contest
     "expected_since_1.4.0.toml": {
         ("Andrews, M.D.", ("given", "suffix")): "fix(comma-family)",
         ("田中, 太郎さん", ("given", "suffix")): "fix(cjk-comma-honorific-peel)",

tools/differential/compare.py

@@ -25,7 +25,7 @@
 FIELDS = ("title", "first", "middle", "last", "suffix", "nickname",
           "maiden")
 
-DEFAULT_BASELINE = "2.1.0"
+DEFAULT_BASELINE = "2.2.0"
 REPO_ROOT = HERE.parents[1]
 #: The v2 API's names for the same seven roles FIELDS names in v1
 #: vocabulary. Both are compared from baseline 2.0 on.

tools/differential/expected_since_2.2.0.toml

@@ -0,0 +1,34 @@
+# Ledger for baseline 2.2.0 -- what changes for a user upgrading from
+# the previous minor. Same rule grammar as the other ledgers: `issue`,
+# `name_regex` and `fields` are every one of them required. Neither
+# narrowing key may stand on its own -- #451 banned a rule with
+# `fields` and no regex, #456 banned the reverse -- so a rule here
+# narrows by name AND by role, never by just one.
+#
+# Opened empty the day 2.2.0 shipped (AGENTS.md release step 8), and it
+# is opened before it has anything to say because DEFAULT_BASELINE and
+# the ledger are coupled: _allowlist_for treats a missing file as a
+# hard error, on the reasoning that an absent ledger classifies nothing
+# and would make every diff report as unexplained. So the baseline
+# could not advance to 2.2.0 without this file.
+#
+# Do not copy rules across from expected_since_2.1.0.toml: that ledger
+# classifies what 2.2 changed on top of 2.1, and a rule copied without
+# checking either over-matches -- hiding a real 2.3 regression behind a
+# 2.2-era label -- or never fires at all. Where a 2.3 change is visible
+# from several baselines the rule may legitimately appear in more than
+# one file, but each copy is checked against its own run rather than
+# assumed.
+#
+# tests/v2/test_differential.py permits exactly one empty ledger, the
+# one DEFAULT_BASELINE names -- which is this file, for as long as the
+# 2.3 cycle moves no corpus name at the default order.
+#
+# tests/v2/test_ledger_guards.py records this file in
+# _SPAN_BEARING_RULES with an empty set and in _CORPUS_CLAIMS with an
+# empty mapping. The first rule that hand-copies _SCRIPT_RANGES has to
+# be recorded in the former or the sweep fails.
+#
+# There is deliberately no `change = []` line. TOML forbids appending a
+# [[change]] table to a statically defined array, so that line would
+# block the first entry.