From bba14cb9ba4de180eac3a091ea51c708fa3b80d4 Mon Sep 17 00:00:00 2001
From: Jeroen Schweitzer <jpmschweitzer@gmail.com>
Date: Thu, 16 Jul 2026 19:05:43 +0200
Subject: [PATCH] test(client): retry input-roundtrip server spawn on fresh
 port when it dies pre-connect

The server binds AFTER ~250ms of plugin/sim init, so a random-port collision (os error 98) escapes the 0.15s spawn-liveness check and surfaced as a hard test failure at the push gate. Death during the connect window now retries on a fresh port within MAX_PORT_ATTEMPTS. Verified: targeted suite 6/6.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 client/tests/test_input_roundtrip.gd | 75 +++++++++++++++-------------
 1 file changed, 41 insertions(+), 34 deletions(-)

diff --git a/client/tests/test_input_roundtrip.gd b/client/tests/test_input_roundtrip.gd
index 71c8136cd..3354a910d 100644
--- a/client/tests/test_input_roundtrip.gd
+++ b/client/tests/test_input_roundtrip.gd
@@ -26,17 +26,13 @@ static func _random_test_port() -> int:
 
 
 func _spawn_server(server_path: String) -> bool:
-	for attempt in range(MAX_PORT_ATTEMPTS):
-		_test_port = _random_test_port()
-		var addr := "127.0.0.1:%d" % _test_port
-		_server_pid = OS.create_process(server_path, [addr])
-		if _server_pid <= 0:
-			continue
-		await get_tree().create_timer(0.15).timeout
-		if OS.is_process_running(_server_pid):
-			return true
+	_test_port = _random_test_port()
+	var addr := "127.0.0.1:%d" % _test_port
+	_server_pid = OS.create_process(server_path, [addr])
+	if _server_pid <= 0:
 		_server_pid = -1
-	return false
+		return false
+	return true
 
 
 func after_test() -> void:
@@ -51,37 +47,48 @@ func after_test() -> void:
 ## Spawn the server, connect, and complete the protocol handshake.
 ## Binary absence is handled by per-test `_do_skip` — by the time this runs the
 ## binary exists, so any failure here is a real failure (asserted loudly).
+## The server binds its port AFTER plugin/simulation init (~250ms), so a
+## random-port collision surfaces as a mid-connect process death, not an
+## instant spawn failure — a death during the connect window therefore
+## retries on a fresh port instead of failing the test (the push gate hit
+## exactly that race: "Failed to bind ... Address already in use").
 func _connect_to_server() -> bool:
 	var server_path := _server_binary_path()
 
-	var spawned := await _spawn_server(server_path)
-	assert_bool(spawned).override_failure_message(
-		"server spawn failed after %d port attempts" % MAX_PORT_ATTEMPTS
-	).is_true()
-	if not spawned:
-		return false
-
-	_bridge = LocalBridge.new()
 	var connected := false
-	var elapsed := 0.0
-	while elapsed < CONNECT_TIMEOUT:
-		if _server_pid > 0 and not OS.is_process_running(_server_pid):
-			push_warning("Server process died during connection")
+	for _attempt in range(MAX_PORT_ATTEMPTS):
+		if not _spawn_server(server_path):
+			continue
+		_bridge = LocalBridge.new()
+		var elapsed := 0.0
+		while elapsed < CONNECT_TIMEOUT:
+			if not OS.is_process_running(_server_pid):
+				push_warning(
+					"Server died pre-connect (port %d likely in use) — retrying" % _test_port
+				)
+				break
+			if _bridge.get_status() == StreamPeerTCP.STATUS_NONE:
+				_bridge.connect_to_server("127.0.0.1", _test_port)
+			_bridge.poll()
+			if _bridge.get_status() == StreamPeerTCP.STATUS_CONNECTED:
+				connected = true
+				break
+			if _bridge.get_status() == StreamPeerTCP.STATUS_ERROR:
+				_bridge.disconnect_from_server()
+				_bridge.reset()
+			await get_tree().create_timer(0.1).timeout
+			elapsed += 0.1
+		if connected:
 			break
-		if _bridge.get_status() == StreamPeerTCP.STATUS_NONE:
-			_bridge.connect_to_server("127.0.0.1", _test_port)
-		_bridge.poll()
-		if _bridge.get_status() == StreamPeerTCP.STATUS_CONNECTED:
-			connected = true
-			break
-		if _bridge.get_status() == StreamPeerTCP.STATUS_ERROR:
-			_bridge.disconnect_from_server()
-			_bridge.reset()
-		await get_tree().create_timer(0.1).timeout
-		elapsed += 0.1
+		_bridge.disconnect_from_server()
+		_bridge = null
+		if _server_pid > 0 and OS.is_process_running(_server_pid):
+			OS.kill(_server_pid)
+		_server_pid = -1
 
 	assert_bool(connected).override_failure_message(
-		"TCP connect to spawned server failed within %.1fs" % CONNECT_TIMEOUT
+		"TCP connect to spawned server failed within %.1fs across %d port attempts"
+		% [CONNECT_TIMEOUT, MAX_PORT_ATTEMPTS]
 	).is_true()
 	if not connected:
 		return false