Unleash / unleash / 21475376490

86%

master: 91%

Build Type

push

github

Committed by web-flow

Commit Message

chore: don't mount a separate router for the conditional middleware (#11268)

Discovered in
https://github.com/bricks-software/unleash-enterprise/pull/762, there's
an issue with using fake timers and the old implementation of the
condititonal middleware.

By changing the implementation to not use a Router, but instead just a
plain function, we can avoid the issue with fake timers.

## AI explanation

According to Claude:

The issue is the interaction between **`vi.useFakeTimers()`** and
**Express `Router`**.

When Vitest (or Jest) fake timers are active, they replace the global
`setTimeout`, `setInterval`, `setImmediate`, and related functions with
mocked versions that only advance when you explicitly tell them to
(e.g., `vi.advanceTimersByTime`).

### Why the Router causes a hang

Express's `Router` internally uses **asynchronous scheduling** —
specifically `setImmediate` or `process.nextTick` in certain code paths
for dispatching requests through its middleware stack. When you wrap
your middleware in a `Router`:

```typescript
const router = Router();
router.use((req, res, next) => {
    // ...
});
return router;
```

The Router's internal dispatch mechanism may schedule work via
`setImmediate`. With fake timers active, **that `setImmediate` never
fires** because it's been replaced by a mock that waits for
`vi.advanceTimersByTime()` or `vi.runAllTimers()`. The HTTP request
enters the Router and never comes back — it just hangs.

### Why a plain function works

A plain handler function:

```typescript
return (req, res, next) => {
    if (condition()) {
        middleware(req, res, next);
    } else {
        next();
    }
};
```

...runs **synchronously inline** in the existing middleware chain.
There's no sub-router dispatch, no `setImmediate`, nothing for fake
timers to intercept. It just calls `next()` directly and everything
proceeds.

Run Details

1692 of 1904 branches covered (88.87%)

1 of 1 new or added line in 1 file covered. (100.0%)

3 existing lines in 2 files now uncovered.

14288 of 16576 relevant lines covered (86.2%)

871.38 hits per line