From 39b1bd03a617132f3a9eb966f8bfc2305611082d Mon Sep 17 00:00:00 2001
From: agra <alex@swipelab.co>
Date: Tue, 19 May 2026 18:57:26 +0300
Subject: [PATCH] issue-0038: closure free-var analysis skips FfiIntrinsicCall
 nodes
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Surfaced while writing the Phase 1.11 in-construct test. The
closure free-variable analyzer doesn't recursively visit the
`ffi_intrinsic_call` AST node introduced in Phase 1.1, so any
identifier used inside `#objc_call` / `#jni_call` /
`#jni_static_call` from a closure body trips:

  error: unresolved: '<name>'

The same identifier captured from the same scope into a plain
expression resolves fine — so the bug is localized to whatever
recursive arm-walk powers the capture analysis.

Likely fix: add an `ffi_intrinsic_call => { ... }` arm wherever
the `.call =>` arm visits `callee` + `args`. Candidate files:
  - src/sema.zig (capture / scope tracking)
  - src/ir/lower.zig (closure body lowering / `lowerLambda`)
Both should be checked.

Workaround in the meantime: reach the captured value via a
module-level global from inside the closure body. See the
`g_hasher_recv` pattern in
examples/ffi-objc-call-09-in-construct.sx for an applied
instance.
---
 examples/issue-0038.sx         | 41 ++++++++++++++++++++++++++++++++++
 tests/expected/issue-0038.exit |  1 +
 tests/expected/issue-0038.txt  |  1 +
 3 files changed, 43 insertions(+)
 create mode 100644 examples/issue-0038.sx
 create mode 100644 tests/expected/issue-0038.exit
 create mode 100644 tests/expected/issue-0038.txt

diff --git a/examples/issue-0038.sx b/examples/issue-0038.sx
new file mode 100644
index 0000000..bd613bf
--- /dev/null
+++ b/examples/issue-0038.sx
@@ -0,0 +1,41 @@
+// Closure capture analysis doesn't trace into the `FfiIntrinsicCall`
+// AST node — identifiers used inside `#objc_call` / `#jni_call` /
+// `#jni_static_call` from a closure body aren't recognized as
+// captured variables. Surfaced when writing `ffi-objc-call-09-in-
+// construct.sx`.
+//
+// Reduced repro: capture in a closure body works fine for
+// "normal" expressions (see `passthrough_works`), but the same
+// capture inside `#objc_call`'s arg list trips
+// "unresolved: 'recv'" (see `passthrough_via_objc_call` — would
+// fail at parse time, so it's commented out).
+//
+// Likely fix: in the closure free-variable analyzer (sema.zig /
+// lower.zig), add a recursive arm for `ffi_intrinsic_call` that
+// visits `return_type` + every `args[i]` the same way the `.call`
+// arm walks `callee` + `args`.
+
+#import "modules/std.sx";
+#import "modules/compiler.sx";
+
+passthrough_works :: (recv: *void) -> Closure(s32) -> *void {
+    closure((d: s32) -> *void => recv);  // captures `recv` — fine
+}
+
+// passthrough_via_objc_call :: (recv: *void) -> Closure(s32) -> s64 {
+//     // Same `recv` capture, but inside `#objc_call(...)`:
+//     //   error: unresolved: 'recv'
+//     closure((d: s32) -> s64 => #objc_call(s64)(recv, "hash"));
+// }
+
+main :: () -> s32 {
+    inline if OS == .macos {
+        f := passthrough_works(null);
+        p := f(0);
+        print("ok (passthrough works) = {}\n", p == null);
+    }
+    inline if OS != .macos {
+        print("skipped (not macos)\n");
+    }
+    0;
+}
diff --git a/tests/expected/issue-0038.exit b/tests/expected/issue-0038.exit
new file mode 100644
index 0000000..573541a
--- /dev/null
+++ b/tests/expected/issue-0038.exit
@@ -0,0 +1 @@
+0
diff --git a/tests/expected/issue-0038.txt b/tests/expected/issue-0038.txt
new file mode 100644
index 0000000..6d77b8c
--- /dev/null
+++ b/tests/expected/issue-0038.txt
@@ -0,0 +1 @@
+ok (passthrough works) = true