test/onboard-selection.test.ts (2)

1256-1271: ⚡ Quick win

Assert the visible repair transcript in the already-loopback path.

This is the silent-skip regression called out in #3337, but the test only checks side effects. Please also assert that this path prints Configuring Ollama systemd loopback override... so the required user-visible confirmation cannot regress.

Suggested assertion

     assert.ok(
       payload.shellCommands.some((command: string) =>
         command.includes("/etc/systemd/system/ollama.service.d/override.conf"),
       ),
       "already-loopback systemd Ollama still needs the persistent drop-in",
     );
+    assert.ok(
+      payload.lines.some((line: string) =>
+        line.includes("Configuring Ollama systemd loopback override"),
+      ),
+      "already-loopback repair should emit the visible loopback-override transcript",
+    );
     assert.ok(
       !payload.shellCommands.some((command: string) =>
         command.includes("OLLAMA_HOST=127.0.0.1:11434 ollama serve"),
       ),

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@test/onboard-selection.test.ts` around lines 1256 - 1271, Add an assertion to
verify the user-visible repair transcript for the already-loopback path: after
the existing assertions that inspect payload.shellCommands and
payload.tagsProbeCount in onboard-selection.test.ts, assert that
payload.transcript (or the equivalent transcript/log field used in this test)
includes the string "Configuring Ollama systemd loopback override..." so the
visible confirmation cannot regress; use an assert.ok check with a clear failure
message referencing the already-loopback path.

---

1196-1216: ⚡ Quick win

Make the re-probe assertion order-aware.

tagsProbeCount >= 2 does not prove a /api/tags probe happened after the systemd restart. Two early probes would still pass. Record an event timeline and assert that at least one tags probe occurs after the restart command.

Suggested tightening

-let tagsProbeCount = 0;
+const events = [];
 const shellCommands = [];
 
 runner.runCapture = (command) => {
   const cmd = Array.isArray(command) ? command.join(" ") : command;
   if (cmd.includes("command -v ollama")) return "/usr/bin/ollama";
   if (cmd.includes("127.0.0.1:11434/api/tags")) {
-    tagsProbeCount += 1;
+    events.push("tags");
     return JSON.stringify({ models: [{ name: "qwen3:8b" }] });
   }
@@
 runner.runShell = (command) => {
   shellCommands.push(command);
+  if (command.includes("sudo systemctl restart ollama")) events.push("restart");
   return { status: 0 };
 };
@@
-    originalLog(JSON.stringify({ result, lines, shellCommands, tagsProbeCount }));
+    originalLog(JSON.stringify({ result, lines, shellCommands, events }));
@@
-    assert.ok(
-      payload.tagsProbeCount >= 2,
-      "should re-probe after the systemd restart instead of trusting a stale loopback cache",
-    );
+    const restartIndex = payload.events.indexOf("restart");
+    assert.ok(restartIndex >= 0, "expected a systemd restart");
+    assert.ok(
+      payload.events.slice(restartIndex + 1).includes("tags"),
+      "should re-probe after the systemd restart instead of trusting a stale loopback cache",
+    );

Also applies to: 1268-1271

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@test/onboard-selection.test.ts` around lines 1196 - 1216, The current test
only counts tags probes via tagsProbeCount, which can’t prove a probe happened
after the restart; change the stubbing in runner.runCapture and runner.runShell
to record ordered events (e.g., push timestamped strings into shellCommands or a
new eventTimeline array inside runner.runCapture when the "/api/tags" branch
runs and inside runner.runShell when the systemd restart command is issued),
keep tagsProbeCount if you want but append entries like "tags-probe" and
"systemctl-restart" to the timeline, then replace the existing assertion
(tagsProbeCount >= 2) with an order-aware assertion that verifies at least one
"tags-probe" event appears after the "systemctl-restart" event in the recorded
timeline; update the analogous checks at 1268-1271 the same way.

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

test/onboard-selection.test.ts (1)

1380-1413: ⚡ Quick win

Also assert that override-apply failure never falls back to ollama serve.

This branch is meant to fail closed, but the test currently only checks stderr text. A regression that starts an unmanaged daemon before throwing would still pass here. Mirroring the manual-start sentinel used in the restart-failure test would close that gap.

Suggested fix

 runner.runShell = (command) => {
+  if (command.includes("ollama serve")) console.error("manual-start");
   if (command.includes("install -D -m 0644")) return { status: 1 };
   return { status: 0 };
 };
@@
     assert.equal(result.status, 1);
     assert.match(result.stderr, /Failed to apply Ollama systemd loopback override/);
     assert.match(result.stderr, /Refusing to continue/);
+    assert.doesNotMatch(result.stderr, /manual-start/);
   });

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@test/onboard-selection.test.ts` around lines 1380 - 1413, The test must also
assert that a failed override-apply never falls back to spawning an unmanaged
Ollama daemon; after running the script that calls setupNim, add an assertion
(mirroring the restart-failure test's "manual-start" sentinel) that the
manual-start sentinel file was NOT created in the test's HOME/tmpDir area so we
guarantee no unmanaged "ollama serve" was started; locate this check alongside
the existing assertions on result and stderr (referencing variables result and
tmpDir and the setupNim invocation).

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes