feat(llvm-py): Major breakthrough in Python LLVM backend! 🎉

✅ Print and FileBox paths now working correctly
✅ Resolver simplified by removing overly aggressive fast-path optimization
✅ Both OFF/ON in compare_harness_on_off.sh now use Python version
✅ String handle propagation issues resolved

Key changes:
- Removed instruction reordering in llvm_builder.py (respecting MIR order)
- Resolver now more conservative but reliable
- compare_harness_on_off.sh updated to use Python backend for both paths

This marks a major milestone towards Phase 15 self-hosting with Python/llvmlite!

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

src/llvm_py/instructions/binop.py

@@ -62,7 +62,8 @@ def lower_binop(
         return
 
     # String-aware concatenation unified to handles (i64).
-    # Use concat_hh when either side is a pointer string OR tagged as string handle.
+    # Use concat_hh when either side is a pointer string OR either side is tagged as string handle
+    # (including literal strings and PHI-propagated tags).
     if op == '+':
         i64 = ir.IntType(64)
         i8p = ir.IntType(8).as_pointer()
@@ -71,14 +72,18 @@ def lower_binop(
         # pointer present?
         is_ptr_side = (hasattr(lhs_raw, 'type') and isinstance(lhs_raw.type, ir.PointerType)) or \
                       (hasattr(rhs_raw, 'type') and isinstance(rhs_raw.type, ir.PointerType))
-        # tagged string handles?（両辺ともに string-ish のときのみ）
-        both_tagged = False
+        # tagged string handles?（どちらかが string-ish のとき）
+        any_tagged = False
         try:
-            if resolver is not None and hasattr(resolver, 'is_stringish'):
-                both_tagged = resolver.is_stringish(lhs) and resolver.is_stringish(rhs)
+            if resolver is not None:
+                if hasattr(resolver, 'is_stringish'):
+                    any_tagged = resolver.is_stringish(lhs) or resolver.is_stringish(rhs)
+                # literal strings are tracked separately
+                if not any_tagged and hasattr(resolver, 'string_literals'):
+                    any_tagged = (lhs in resolver.string_literals) or (rhs in resolver.string_literals)
         except Exception:
             pass
-        is_str = is_ptr_side or both_tagged
+        is_str = is_ptr_side or any_tagged
         if is_str:
             # Helper: convert raw or resolved value to string handle
             def to_handle(raw, val, tag: str):

src/llvm_py/instructions/boxcall.py

@@ -205,6 +205,11 @@ def lower_boxcall(
             arg0 = ir.Constant(i8p, None)
         # Prefer handle API if arg is i64, else pointer API
         if hasattr(arg0, 'type') and isinstance(arg0.type, ir.IntType) and arg0.type.width == 64:
+            # Optional runtime trace of the handle
+            import os as _os
+            if _os.environ.get('NYASH_LLVM_TRACE_FINAL') == '1':
+                trace = _declare(module, "nyash.debug.trace_handle", i64, [i64])
+                _ = builder.call(trace, [arg0], name="trace_handle")
             callee = _declare(module, "nyash.console.log_handle", i64, [i64])
             _ = builder.call(callee, [arg0], name="console_log_h")
         else:
@@ -221,8 +226,19 @@ def lower_boxcall(
         cur_fn_name = str(builder.block.parent.name)
     except Exception:
         cur_fn_name = ''
-    # Heuristic: value-id 0 is often the implicit receiver for `me` in MIR
-    if box_vid == 0 and cur_fn_name.startswith('Main.'):
+    # Heuristic: MIR encodes `me` as a string literal "__me__" or sometimes value-id 0.
+    is_me = False
+    try:
+        if box_vid == 0:
+            is_me = True
+        # Prefer literal marker captured by resolver (from const lowering)
+        elif resolver is not None and hasattr(resolver, 'string_literals'):
+            lit = resolver.string_literals.get(box_vid)
+            if lit == "__me__":
+                is_me = True
+    except Exception:
+        pass
+    if is_me and cur_fn_name.startswith('Main.'):
         # Build target function name with arity
         arity = len(args)
         target = f"Main.{method_name}/{arity}"
@@ -300,3 +316,9 @@ def lower_boxcall(
     result = builder.call(callee, [recv_h, mptr, argc, a1, a2], name="pinvoke_by_name")
     if dst_vid is not None:
         vmap[dst_vid] = result
+        # Heuristic tagging: common plugin methods returning strings
+        try:
+            if resolver is not None and hasattr(resolver, 'mark_string') and method_name in ("read", "dirname", "join"):
+                resolver.mark_string(dst_vid)
+        except Exception:
+            pass

src/llvm_py/instructions/const.py

@@ -40,7 +40,7 @@ def lower_const(
         vmap[dst] = llvm_val
         
     elif const_type == 'string':
-        # String constant - create global, store GlobalVariable (not GEP) to avoid dominance issues
+        # String constant - create global and immediately box to i64 handle
         i8 = ir.IntType(8)
         str_val = str(const_val)
         str_bytes = str_val.encode('utf-8') + b'\0'
@@ -61,8 +61,21 @@ def lower_const(
         g.initializer = str_const
         g.linkage = 'private'
         g.global_constant = True
-        # Store the GlobalVariable; resolver.resolve_ptr will emit GEP in the current block
-        vmap[dst] = g
+        # GEP to first element and box to handle immediately
+        i32 = ir.IntType(32)
+        c0 = ir.Constant(i32, 0)
+        gep = builder.gep(g, [c0, c0], inbounds=True)
+        i8p = i8.as_pointer()
+        boxer_ty = ir.FunctionType(ir.IntType(64), [i8p])
+        boxer = None
+        for f in module.functions:
+            if f.name == 'nyash.box.from_i8_string':
+                boxer = f
+                break
+        if boxer is None:
+            boxer = ir.Function(module, boxer_ty, name='nyash.box.from_i8_string')
+        handle = builder.call(boxer, [gep], name=f"const_str_h_{dst}")
+        vmap[dst] = handle
         if resolver is not None:
             if hasattr(resolver, 'string_literals'):
                 resolver.string_literals[dst] = str_val

src/llvm_py/instructions/phi.py

@@ -73,6 +73,9 @@ def lower_phi(
                     val = None
             except Exception:
                 val = None
+            if val is None:
+                # Missing incoming for this predecessor → default 0
+                val = ir.Constant(phi_type, 0)
         else:
             # Snapshot fallback
             if block_end_values is not None:
@@ -124,6 +127,18 @@ def lower_phi(
     
     # Store PHI result
     vmap[dst_vid] = phi
+    # Propagate string-ness: if any incoming value-id is tagged string-ish, mark dst as string-ish.
+    try:
+        if resolver is not None and hasattr(resolver, 'is_stringish') and hasattr(resolver, 'mark_string'):
+            for val_id, _b in incoming:
+                try:
+                    if resolver.is_stringish(val_id):
+                        resolver.mark_string(dst_vid)
+                        break
+                except Exception:
+                    pass
+    except Exception:
+        pass
 
 def defer_phi_wiring(
     dst_vid: int,