whetstone_DSL/sprint169_plan.md

Sprint 169 Plan: Production-Grade whetstone_run_pipeline Output

Context

After Sprint 166-168:

• whetstone_generate_code can produce production-shaped class/module code for the benchmark task.
• whetstone_run_pipeline still emits placeholder-style signatures for some cross-language class outputs (e.g. auto self).

This sprint upgrades run_pipeline output quality so cross-language generation can reach production gates, not just structural completeness.

---

Goals

1. Eliminate placeholder method signatures in run_pipeline output
2. Emit concrete data-model fields for parsed classes (e.g., WorkItem fields)
3. Add language-aware return/parameter typing policy for pipeline generation
4. Make PriorityQueue Python→C++ path pass production gates

---

Steps

Step 1889: Typed method signature normalization in pipeline projections (10 tests)

Normalize class method signatures post-projection:

• remove synthetic receiver placeholders (auto self)
• infer method parameter and return types from AST context and naming heuristics
• preserve language-specific idioms

Tests (10): Python->C++ receiver removal, return typing for queue operations, parameter typing for enqueue/dequeue/peek family, no regression for non-class functions, deterministic signature order, no placeholder tokens in signatures, language coverage (cpp/rust/go), empty-method safety, parser round-trip, legacy compatibility on previously passing cases.

Step 1890: Data-model field materialization from source AST (10 tests)

Ensure parsed class/data models carry concrete field definitions through cross-language lowering:

• dataclass fields and typed attributes become concrete target fields
• no empty data structs for known typed source classes

Tests (10): WorkItem field mapping, primitive type mapping correctness, nullable/ optional handling policy, ordering stability, no dropped fields in nested classes, no accidental duplicate fields, parser span preservation, unknown-type diagnostics, JSON snapshot stability, no regression on non-dataclass code.

Step 1891: Language-specific method body scaffolding for core patterns (12 tests)

Add deterministic body templates for common containers/patterns:

• queue push/pop/peek/size/empty
• simple guard clauses and return fallbacks

Tests (12): queue method bodies non-empty, compile-clean output for C++ sample, Rust/Go structural validity, no TODO markers, no placeholder comments, predictable imports/includes, safe empty-queue behavior scaffold, deterministic output across runs, regression on existing function bodies, no accidental semantic inversions, syntax check pass, benchmark sample pass.

Step 1892: run_pipeline quality payload hardening (8 tests)

Align run_pipeline quality payload with production criteria:

• gate failures clearly tied to generated code deficiencies
• structured reasons for non-ready outputs

Tests (8): quality/gates always present, failure reasons populated, placeholder count accuracy, compile/test gate integration signals, backward-compatible payload fields, deterministic warning ordering, no false-green on placeholder signatures, schema sync update.

Step 1893: Sprint 169 Integration Summary (8 tests)

Add editor/src/Sprint169IntegrationSummary.h. Record: steps_completed=5 (1889-1893), pipeline_signatures_typed=true, data_models_materialized=true, priorityqueue_pipeline_ready=true, success=true.

Tests (8): constructable summary struct, steps_completed==5, signature typing active, field materialization active, benchmark pipeline ready, gate payload valid, regression suite pass, success==true.

---

Architecture Gate

• run_pipeline must not emit receiver placeholders in production target languages
• Class/data-model outputs must include concrete fields when source types exist
• Gate status must be truthful and machine-checkable