From b502f494e2dda5b26459c2a6a9af81c8645c668a Mon Sep 17 00:00:00 2001
From: hellices <hellices@naver.com>
Date: Sat, 24 Oct 2026 02:12:50 +0900
Subject: [PATCH] test: wait for an approval dialog to compose before reading
 it

The operation-journey driver read `.confirm-title` as soon as app.screen
was a ConfirmScreen. Textual makes a pushed screen current before it
composes the screen's widgets, so on a loaded runner the poll landed in
between and the read raised "No nodes match '.confirm-title' on
ConfirmScreen()". That failed two journeys on main's 3.11 (#478); the
second NoMatches in that log came from the journey tearing down after
the first.

handle() now first waits, through _await_dialog_composed, until the
current dialog has its title or is no longer current; a dialog that
closes first falls through to the existing closed-before-a-decision
path. The new test delays ConfirmScreen's compose by 0.3 seconds and
still expects one matched dialog_observed and no hard failures; without
the wait it fails every run with CI's exact pair of errors.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---
 tests/evals/operation_app.py           | 18 ++++++++++++++++++
 tests/evals/test_operation_journeys.py | 21 +++++++++++++++++++++
 2 files changed, 39 insertions(+)

diff --git a/tests/evals/operation_app.py b/tests/evals/operation_app.py
index 02f5d0d5..93b4d3ab 100644
--- a/tests/evals/operation_app.py
+++ b/tests/evals/operation_app.py
@@ -725,7 +725,25 @@ def _apply_intervention(self) -> None:
                 f"{self._journey.id}: dialog_intervention replacement did not apply"
             )
 
+    async def _await_dialog_composed(self, pilot: Any) -> None:
+        """Wait until a current approval dialog has its widgets, or closes.
+
+        Textual makes a pushed screen current before it composes the screen's
+        widgets, and a loaded runner can poll in between (#478): reading the
+        title then would find none. A dialog that closes first is left to
+        `handle`, which already treats it as closed before a decision.
+        """
+        screen = self._app.screen
+        if not isinstance(screen, ConfirmScreen):
+            return
+        await until(
+            pilot,
+            lambda: self._app.screen is not screen or bool(screen.query(".confirm-title")),
+            label="approval dialog composed",
+        )
+
     async def handle(self, pilot: Any) -> None:
+        await self._await_dialog_composed(pilot)
         screen = self._app.screen
         if not isinstance(screen, ConfirmScreen):
             if self._remaining <= 0 or not self._dialog_seen:
diff --git a/tests/evals/test_operation_journeys.py b/tests/evals/test_operation_journeys.py
index 54e97d94..7ba9d569 100644
--- a/tests/evals/test_operation_journeys.py
+++ b/tests/evals/test_operation_journeys.py
@@ -8,6 +8,7 @@
 
 from __future__ import annotations
 
+import asyncio
 import json
 import os
 from collections.abc import AsyncIterator, Callable
@@ -31,6 +32,7 @@
 from korvid.evals.scripted import ScriptedProvider
 from korvid.tools.executor import RecordedExecution, ToolOutcome
 from korvid.tools.structured import dump_yaml
+from korvid.ui.widgets.confirm_screen import ConfirmScreen
 
 from . import operation_app
 from .operation_app import (
@@ -1331,6 +1333,25 @@ def polls_that_miss_every_dialog(
     assert run.audit == ()
 
 
+async def test_a_dialog_still_composing_is_read_once_it_has_its_widgets(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """Textual makes a pushed screen current before it composes the screen's
+    widgets, and a loaded runner can poll in between (#478). The driver must
+    wait for the dialog's title instead of reading a dialog that has none."""
+    original = ConfirmScreen._compose
+
+    async def slow_compose(self: ConfirmScreen) -> None:
+        await asyncio.sleep(0.3)
+        await original(self)
+
+    monkeypatch.setattr(ConfirmScreen, "_compose", slow_compose)
+    run = await run_scripted_journey("scale-deployment-up", tmp_path)
+    observed = [entry for entry in run.journal if entry["event"] == "dialog_observed"]
+    assert [entry["result"] for entry in observed] == ["matched"]
+    assert run.grade.hard_failures == ()
+
+
 def test_the_replacement_journey_declares_its_intervention_in_the_fixture() -> None:
     """B1: the mid-dialog replacement is fixture data, not test code, so
     `tests.evals.operation_campaign` runs the identical journey."""