40 lines
1.4 KiB
Markdown
40 lines
1.4 KiB
Markdown
# Sprint 143 Plan: AST-to-Text Deterministic Regeneration with C++ Style Guards
|
|
|
|
## Context
|
|
|
|
Hybrid authoring requires AST changes to regenerate readable, stable C++ source.
|
|
Sprint 143 adds deterministic regeneration with style/format guardrails.
|
|
|
|
---
|
|
|
|
## Goals
|
|
|
|
1. Regenerate C++ text from AST with stable formatting
|
|
2. Preserve user style profile and ordering where policy allows
|
|
3. Minimize noisy diffs from semantically neutral changes
|
|
4. Expose regeneration decisions for auditability
|
|
|
|
---
|
|
|
|
## Steps
|
|
|
|
### Step 1669: AST-to-text regeneration policy schema (10 tests)
|
|
### Step 1670: C++ formatting/styling guard model (10 tests)
|
|
### Step 1671: Stable include/order regeneration model (10 tests)
|
|
### Step 1672: Semantic-noise diff suppressor model (10 tests)
|
|
### Step 1673: `whetstone_regenerate_text_from_ast` MCP tool (8 tests)
|
|
### Step 1674: `whetstone_preview_regenerated_diff` MCP tool (8 tests)
|
|
### Step 1675: `whetstone_get_regeneration_decisions` MCP tool (8 tests)
|
|
### Step 1676: Regeneration quality scorer model (8 tests)
|
|
### Step 1677: Regeneration stability report artifact (8 tests)
|
|
### Step 1678: Sprint 143 integration summary + regression (8 tests)
|
|
|
|
---
|
|
|
|
## Architecture Gate
|
|
|
|
- Regeneration must be deterministic and reproducible
|
|
- Style guard output must avoid non-semantic churn
|
|
- Decision logs must be structured and machine-readable
|
|
- Max 600 lines per header
|