Add KeepAlive instruction to fix hidden root problem where x = null
doesn't properly drop the strong reference.
Key changes:
- Add KeepAlive { values, drop_after } instruction to MIR
- Emit KeepAlive[drop=true] in build_assignment() before variable overwrite
- Emit KeepAlive[drop=false] in pop_lexical_scope() at scope end
- VM handler: when drop_after=true, remove ALL ValueIds pointing to
the same Arc (handles SSA Copy chains)
Test results:
- weak_upgrade_fail: exit 1 ✅ (weak_to_strong returns null after x=null)
- weak_basic: exit 2 ✅ (weak_to_strong succeeds while x alive)
- quick smoke: 154/154 PASS ✅🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Fix LLVM print to output 42 instead of 4 (handle value) for field access.
Root cause: Type tags lost through MIR copy instruction chains
- getField tagged ValueId 16 as handle
- MIR copy chain: 16 → 17 → 18
- print used ValueId 18 (not tagged) → treated as raw integer
Solution: Type-tag based handle detection with copy propagation
- boxcall.py: Tag getField results as handles
- global_call.py: Skip boxing for handles in print
- copy.py: Propagate value_types tags through copy chains
Test coverage:
- apps/tests/phase285_print_raw_int.hako: Raw int regression check
- apps/tests/phase285_userbox_field_basic.hako: Field access parity
Result: VM/LLVM parity achieved (both output 42) ✅🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Remove automatic WeakNew conversion and enforce strict compile-time type
checking for weak field assignments. Only 3 assignment types allowed:
1. Result of weak(x) call (WeakRef type)
2. Existing WeakRef variable (e.g., me.parent = other.parent)
3. Void/null (clear operation)
**Implementation**:
- Added MirType::WeakRef to type system (src/mir/types.rs)
- Track WeakRef type in emit_weak_new() even in pure mode
- Weak field reads return WeakRef without auto-upgrade
- Removed automatic WeakNew conversion from field writes
- Implemented check_weak_field_assignment() with actionable errors
- Fixed null literal type tracking (Phase 285A1.1: Unknown → Void)
**Testing**:
- 5 test fixtures (3 OK, 2 NG cases) - all passing
- Smoke test: phase285_weak_field_vm.sh
- Error messages guide users to use weak() or null
**Documentation**:
- Updated lifecycle.md SSOT with weak field contract
🤖 Generated with Claude Code
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
## Summary
Implemented fail-fast validation for PHI ordering and value resolution in strict mode.
## Changes
### P1-1: Strict mode for "PHI after terminator"
- File: `src/llvm_py/phi_wiring/wiring.py::ensure_phi`
- Behavior: `NYASH_LLVM_PHI_STRICT=1` → RuntimeError if PHI created after terminator
- Default: Warning only (no regression)
### P1-2: Strict mode for "fallback 0"
- File: `src/llvm_py/phi_wiring/wiring.py::wire_incomings`
- Behavior: Strict mode forbids silent fallback to 0 (2 locations)
- Location 1: Unresolvable incoming value
- Location 2: Type coercion failure
- Error messages point to next debug file: `llvm_builder.py::_value_at_end_i64`
### P1-3: Connect verify_phi_ordering() to execution path
- File: `src/llvm_py/builders/function_lower.py`
- Behavior: Verify PHI ordering after all instructions emitted
- Debug mode: Shows "✅ All N blocks have correct PHI ordering"
- Strict mode: Raises RuntimeError with block list if violations found
## Testing
✅ Test 1: strict=OFF - passes without errors
✅ Test 2: strict=ON - passes without errors (no violations in test fixtures)
✅ Test 3: debug mode - verify_phi_ordering() connected and running
## Scope
- LLVM harness (Python) changes only
- No new environment variables (uses existing 3 from Phase 277 P2)
- No JoinIR/Rust changes (root fix is Phase 279)
- Default behavior unchanged (strict mode opt-in)
## Next Steps
- Phase 278: Remove deprecated env var support
- Phase 279: Root fix - unify "2本のコンパイラ" pipelines
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Phase 269 P1.0: Pattern8 SSA correctness
- Add PHI node for loop variable `i` in pattern8_scan_bool_predicate.rs
- Ensure proper SSA form: i_current = phi [(preheader, i_init), (step, i_next)]
- Create loop_predicate_scan.rs for Pattern8 Frag emission
- Pre-allocate PHI destination before block generation
- Use insert_phi_at_head_spanned() for span synchronization
Phase 269 P1.1: call_method return type SSOT propagation
- Add callee_sig_name() helper in annotation.rs for function name formatting
- Annotate call_method return types in emit_unified_call_impl()
- Use module signature as SSOT for return type resolution (no hardcoding)
- Arity-aware function name: "BoxName.method/arity"
Fixes: is_integer() now correctly returns Bool instead of String
Test: Simple call_method test returns exit=7 (loop test has pre-existing bug)
Unit tests: All 1389 tests passing
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Problem:
- Pattern3 heuristic was too conservative: detect_if_in_body() returned
true for ANY if statement, causing simple conditional assignments to be
misclassified as Pattern3IfPhi
- Example: `if i == 0 then seg = "first" else seg = "other"` was routed
to Pattern3, but Pattern3 only handles if-sum patterns like
`sum = sum + (if x then 1 else 0)`
- This caused loops with conditional assignment to fail Pattern3 check
and exhaust all routing paths
Solution (Conservative, Phase 264 P0):
- ast_feature_extractor.rs:
- detect_if_else_phi_in_body(): Always return false
- has_if = has_if_else_phi (don't use detect_if_in_body())
- loop_pattern_detection/mod.rs:
- Add has_if_sum_signature() (returns false for P0)
- has_if_else_phi = carrier_count > 1 && has_if_sum_signature(scope)
Effect:
- Simple conditional assignment loops now fall through to Pattern1 ✅
- Pattern3 misrouting prevented ✅
Results:
- Lib tests: 1368/1368 PASS (no regression)
- Minimal repro: phase264_p0_bundle_resolver_loop_min.hako PASS ✅
- Quick smoke: 45/46 (unchanged - complex BundleResolver loop needs P1)
Phase 264 P1 TODO:
- Implement accurate if-sum signature detection (AST/CFG analysis)
- Support complex nested loops in Pattern2 or new pattern
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Pattern8 (Boolean Predicate Scan) implementation for is_integer/1:
- New pattern detection for `loop + if not predicate() { return false }`
- JoinIR lowerer with main/loop_step/k_exit structure
- Me receiver passed as param (by-name 禁止)
Key fixes:
1. expr_result = Some(join_exit_value) (Pattern7 style)
2. Tail-call: dst: None (no extra Ret instruction)
3. instruction_rewriter: Add `&& is_loop_header_with_phi` check
- Pattern8 has no carriers → no PHIs → MUST generate Copy bindings
- Without this, ValueId(103/104/105) were undefined
Status: Copy instructions now generated correctly, but exit block
creation issue remains (next step: Step A-C in指示書).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
**Status**: Core carrier PHI issue partially resolved, debugging loop body
**Progress**:
✅ Task 1: split_scan_minimal.rs Carriers-First ordering (6 locations)
✅ Task 2: pattern7_split_scan.rs boundary configuration (host/join inputs, exit_bindings, expr_result)
✅ Result now flows from k_exit to post-loop code (RC issue resolved)
⚠️ Loop body instruction execution needs review
**Key Fixes**:
1. Fixed host_inputs/join_inputs to match main() params Carriers-First order
2. Added result to exit_bindings (CarrierRole::LoopState)
3. Added result back to loop_invariants for variable initialization
4. Added expr_result=join_exit_value_result for loop expression return
5. Fixed jump args to k_exit to include all 4 params [i, start, result, s]
**Current Issue**:
- Loop body type errors resolved (String vs Integer fixed)
- New issue: Loop body computations (sep_len) undefined in certain blocks
- Likely cause: JoinIR→MIR conversion of local variables needs review
**Next Steps**:
- Review JoinValueSpace allocation and ValueId mapping in conversion
- Verify loop_step instruction ordering and block structure
- May need to refactor bound computation or revisit split algorithm
🤖 Generated with Claude Code
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
