Cleanup

docs/composer/active_path_audit.md

@@ -1,20 +1,25 @@
-# Active solo path after the rewrite
+# Active path audit
 
-The replaced entry path was `production.generate_role → composer.compose → leads.thematic_options → themes.phrase_candidates/unit_candidates → routes/gestures → realization.checked_phrase → scoring.musical`. The same modules retained paraphrase, draft, cell-search, join-refinement, and older scoring APIs. Source-derived feature contracts could preserve a long head rhythm before a new tune existed.
+Starting revision: `0e68f53` (`Cleanup`). The earlier implementation report is preserved in [baseline_0e68f53.md](baseline_0e68f53.md); its test and timing results do not describe this change.
 
-The sole normal path is now `production.generate_role → composer.compose → construction.Builder → rhythm.Rhythms + routes.Routes → scoring.evaluate → placement.alternatives/place → composer.validate → notation/export`. There is no legacy mode, silent fallback, pair selector, or reference argument.
+Before: `production.generate_role → composer.compose → construction.Builder.propose → rhythm.Rhythms.realize + routes.Routes.realize → scoring.evaluate → one-phrase lookahead → placement → validation → notation/export`.
 
-| Finding | Before this rewrite | Resolution |
+After: `production.generate_role → composer.compose → construction.Builder → operators.Operators + selection.Selector → bounded action repair → whole-solo placement → validation → notation/export`.
+
+| Finding | Verified starting behavior | Change / remaining limitation |
 |---|---|---|
-| Multiple generations of musical construction coexist | Confirmed by callers/imports and legacy mode branch | Thirteen obsolete construction/helper modules removed |
-| Mandatory source-feature influence | Confirmed in theme seed and feature-prefix construction | New melody is independent of head rhythm; head references are optional actions |
-| Final scorer overwritten / density omitted | Already repaired in the preceding active path; not reproduced as a current defect | Old duplicate evaluators retired; one completed-phrase evaluator |
-| Broad score often leaves fingering as decisive term | Observed in saved selected/alternative reports | No fingering term in musical selection; whole-melody placement runs after selection |
-| Expression only added after strict selection | Already partly repaired by joint-arrival proposals | All retained production expression construction now precedes route selection; post-passes removed |
-| Bar/cell ends act as mandatory cadences | Older paths contained such requirements; active path had some open behavior | No cells/plans/drafts with competing ending semantics; only actual phrase destinations |
-| Rhythm packets do not establish a recurring idea | Constructor selected separate short material repeatedly | Repeat a whole rhythm group before reshaping it; never slice tuplets |
-| Fixed ending/return counts | Historical settings remained | Those settings and scorers removed; no target counts |
+| Alternatives chosen by hashes before evaluation | `propose` hash-selected timing, destination, treatment, register and first pitch; 32 proposals per batch | Enumerate shared actions before contextual ranking. Stable content ordering breaks score ties. |
+| Complete target cannot pass allocation | Autumn acoustic lead-in was outside export; electric tail clipped before returning vocal | Source pickup measure allocated before composition. Full electric tail allowed with authorized vocal overlap; vocal unchanged. |
+| Scorer overwritten / density silently omitted | Not reproduced: only four named phrase terms remained | One rational action evaluator; no hidden density setting. |
+| Every unit closes | Not reproduced: old constructor had open alternatives, but handoff forced a 2–3 beat hold | Individual actions need no chord-tone stop; handoff can be short or silent. |
+| Answer unreachable | Not reproduced: old chooser requested answers | New answer transforms the actual heard group as a complete candidate. |
+| Expressions are post-processing only | Not reproduced: old proposals planned them jointly | Keep joint treatments; carry pending exact-target obligations across actions. |
+| Missing expressive coverage | Reproduced against actual enumerated operators | General delayed approaches, reattacked anticipations, anticipated colors and separated enclosures. |
+| Color octaves disappear | Midpoint rounding skipped alternating F-sharp octaves | Enumerate explicit legal octaves. |
+| Repeated catalog parsing | Parent parsed catalogs that independent workers parsed again | Parent parse removed; content-addressed immutable JSON catalogs shared by workers. |
 
-Search limits are construction budgets, not musical quotas: 32 proposals per state, eight diverse finalists, two endpoint fingerings per melody, four candidates considered with one-phrase lookahead, one additional planning batch if the first yields nothing. A whole-solo hand-position pass resolves transitions without changing notes.
+`Operators.at` consumes source harmony, physical domain and versioned data. It records proposed, rejected-by-rule and valid counts. `Builder.candidates` retains 24 representatives by kind/duration; pending obligations filter candidates before this cutoff. `Builder.compose` ranks complete actions and allows 16 deterministic backtracks. There is no wall-clock winner, reference-reader import or old-constructor fallback.
 
-Shared source analysis contains imported sounding events, tempo and exact measure boundaries. It does not extract compulsory head motifs or source phrase segmentation. Both roles receive the same physical domain and independent deterministic decision streams.
+This is a local policy, not a globally optimized complete-phrase model. An all-onset frontier experiment was abandoned because of excessive work. Phrase objects group selected actions for validation/reporting; that grouping does not mean whole-phrase alternatives competed. This limitation matters to exact selection.
+
+The development-only `audition.fit_selection` constrains the same enumerator to an explicit target to diagnose representability. Its fitter rolls in with ordinary ranking and updates shared coefficients at the first mismatch. A valid derivation is not proof that production selects it. The release comparison checks the exported unrestricted winner separately.