🎉 This should now be resolved, please upgrade to the latest versions of supertest and superagent:

• supertest v7.1.2 released to npm https://github.com/forwardemail/supertest/releases/tag/v7.1.2
• superagent v10.2.2 released to npm https://github.com/forwardemail/superagent/releases/tag/v10.2.2

🔋 Maintenance supported by Forward Email (@forwardemail) at https://forwardemail.net

Note: I realize the change was not actually introduced here, but commenting here because this issue is referenced in the release notes.

Hey, just wanted to provide some feedback on this change. we had a few test suites which broke when we updated to the latest supertest version. The reason of this failure was that our test suites were designed this way:

let server;

beforeAll(async () => {
  server = app.listen(0);
  await events.once(server, 'listening');
});

afterAll(async () => {
  if (server) {
    await util.promisify(server.close.bind(server)();
  }
});

// a few tests that use server

Previously supertest would not automatically close our tests, because of that we designed our tests to manually close the server. When we upgraded to latest supertest (a patch version), suddenly our test suite always failed with:

⎯⎯⎯⎯⎯⎯ Failed Suites 1 ⎯⎯⎯⎯⎯⎯⎯
 FAIL  test/someTest.js > test name
Error: Server is not running.
⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯[1/1]⎯

 Test Files  1 failed | 6 passed (7)
      Tests  74 passed (74)
   Start at  14:45:40
   Duration  2.42s (transform 196ms, setup 0ms, collect 2.19s, tests 1.66s, environment 2ms, prepare 3.70s)

 ELIFECYCLE  Command failed with exit code 1.

We tracked down that failure to this change. In my opinion, supertest should not close down a server that it does not own, but if we truly want to change this behavior, then I would suggest to clearly document it as a change in behavior, and ideally released as a breaking change instead.

You are correct, let me fix.

Fixed. All tests are passing https://github.com/forwardemail/supertest/releases/tag/v7.1.3

I've already deprecated v7.1.2 as well since it had that backwards incompatible issue and was a patch bump not major