# Sprint 48 Plan: C++ Semantic Raising from Core IR

## Context

Sprint 48 implements the first production target backend for the new IR: modern
C++. The goal is to generate reviewable, idiomatic C++ while preserving behavior
contracts and memory safety intent from Rust lowering.

---

## Goals

1. Raise IR ownership semantics into RAII-centric C++
2. Generate idiomatic type, error, and algorithmic constructs
3. Produce compilable project artifacts (headers, sources, CMake)
4. Emit explicit risk annotations where semantics are weaker than Rust

---

## Steps

### Step 709: Ownership mapping policy engine (`unique_ptr/shared_ptr/value`) (12 tests)
### Step 710: Borrow semantics to references/views strategy (10 tests)
### Step 711: Trait-to-interface/composition raising (10 tests)
### Step 712: Generics/template raising policy (10 tests)
### Step 713: Error model mapping (`Result` policy profiles) (10 tests)
### Step 714: Async mapping strategy (coroutines/task runtime profiles) (10 tests)
### Step 715: STL algorithm lifting from IR intent tags (8 tests)
### Step 716: Build artifact generator (CMake targets + layout) (8 tests)
### Step 717: `whetstone_generate_cpp_from_ir` MCP tool (8 tests)
### Step 718: Sprint 48 integration summary + regression (8 tests)

---

## Output Profiles

- `safe-first`: explicit checks, conservative ownership, clearer diagnostics
- `perf-first`: low-overhead defaults, reduced abstraction layers
- `interop-first`: ABI-friendly shapes for mixed-language systems