test(regression): promote run-flow.spec.ts to @stable

[daniellicnerski1]

Summary

Promotes `tests/tests-automations/regression/flow-functionality/run-flow.spec.ts` to `@stable` after fixing a no-op assertion, trimming excessive timeouts, adding API-based cleanup, and correcting a stale QA-CHECKLIST entry.

Why this spec is non-redundant

This is the only test covering the Run Flow component on canvas — a Langflow component that invokes another flow from within a flow.

• A pipeline flow (ChatInput → TextOutput → ChatOutput) is built via drag-drop
• A caller flow uses the Run Flow component to invoke the pipeline
• The test verifies the input text echoes back through the chain

`api-run-flow.spec.ts` (`@stable`) covers `POST /api/v1/run/{flow_id}` directly via API — different surface. No other UI spec covers flow chaining via the Run Flow component.

Changes

Change	Rationale
Add `@stable` + `@regression` to the tag list	Promotion + accurate functional area
Trim 4x `timeout: 100000` to `timeout: 30000`	100s was excessive copy/paste; project-wide standard is 5-30s
Fix L80 no-op: `await page.getByText("New Flow").isVisible();` → `await expect(...).toBeVisible({ timeout })`	The return value of `.isVisible()` was discarded — the line always passed regardless of actual visibility. Now it actually asserts
Wrap test body in `try/finally` with API cleanup of 2 created flows	Previously left 2 flows orphaned every run. Uses `getAuthToken` + `DELETE /api/v1/flows/{id}` for speed/robustness. Best-effort (errors swallowed) so test failures still report original cause
ESLint --fix: `.toBe()` on inputValue → `.toHaveValue()` auto-waiting	Playwright-native auto-waiting assertion
Fix QA-CHECKLIST L589	Two errors: (1) stale path `core/features/run-flow.spec.ts` (doesn't exist), (2) inaccurate description "Execute flow via Run button" — the spec tests the component, not the canvas Run button

Test plan

• `npm run typecheck` clean
• `npm run lint` filtered on spec: 0 errors (11 warnings: 9 pre-existing `waitForSelector` + 2 intentional conditionals in dotenv check and cleanup)
• Anti-pattern grep — zero `.isVisible().catch(() => false)` / `toBeFalsy` / `waitForTimeout` in test logic
• `npx playwright test --workers=1 --retries=0` — 1 test PASS in 16.1s without retry
• Force-fail confirmed (changed final `expect(...).toHaveValue(...)` to an impossible string; saw failure at L134 with clear message `unexpected value "THIS IS A TEST FOR RUN FLOW COMPONENT"`; reverted; repassed in 18.5s)
• `--trace=on` once — trace shows both flow constructions, execution, and `finally` cleanup with API DELETE calls
• Zero `🚨 Backend Error` occurrences

Known limitation (out of scope)

The cleanup deletes the 2 most-recently-created flows from the listing, assuming reverse-chronological ordering. If a prior orphan from a different test appears newer than the 2 created here, cleanup could delete the wrong flow. Acceptable trade-off — cleanup is best-effort hygiene, and Langflow's listing API correctly orders by recency.

[Copilot]

Pull request overview

Promotes the run-flow.spec.ts regression test to @stable after fixing a no-op visibility assertion, trimming oversized timeouts, adding API-based cleanup of the two flows it creates, and correcting a stale entry in QA-CHECKLIST.

Changes:

• Adds @stable / @regression tags and tightens four 100s timeouts down to 30s; replaces .isVisible() no-op with await expect(...).toBeVisible() and switches the final input assertion to auto-waiting toHaveValue.
• Wraps the test body in try/finally and best-effort deletes the 2 most recently created flows via getAuthToken + DELETE /api/v1/flows/{id}.
• Updates QA-CHECKLIST entry 12.6 to mark the spec stable and corrects its path/description.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
tests/tests-automations/regression/flow-functionality/run-flow.spec.ts	Promote to @stable, trim timeouts, fix no-op assertion, add API cleanup in finally.
QA-CHECKLIST.md	Mark Run Flow component scenario as [x] and fix stale path/description.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[daniellicnerski1]

Valid — fixed in b89103e. You're right that the object-form fallback used the wrong key (body?.items vs the actual body.flows shape established by tests/helpers/flows/clean-all-flows.ts:33). The bug never bit because Langflow currently returns the array form on this endpoint, but a future change to the response shape would have silently skipped cleanup.

The whole listing call is gone now — replaced with the same ID-scoping pattern landed in PR #267 (bulk-actions): the test captures the flow ID from page.url() after each blank-flow.click() and the finally block iterates only those captured IDs. No more list-then-slice; no more object-form ambiguity.

[daniellicnerski1]

Valid — fixed in b89103e. Same root issue as comment 1 (positional slice(0, 2) over the listing). Two concrete risks the previous code had:

1. The list did not use remove_example_flows=true, so any example/starter flow that the API returns ahead of user-created flows would have been deleted.
2. Under fullyParallel: true, a sibling worker could have created flows that ended up in the top-2 positions during the window between request.get(...) and request.delete(...), taking collateral damage.

Fix is the same pattern from PR #267:

• Capture createdFlowIds: string[] populated by page.waitForURL(/\/flow\/[0-9a-f-]+/i) + regex extraction after each blank-flow.click().
• finally block iterates only those IDs and calls DELETE /api/v1/flows/{id} for each. 404s for IDs the test already deleted on the happy path are silenced.

Validated against the live API: 182 flows before the run, 182 after — the 2 captured IDs were deleted, nothing else was touched.