test_runner - using node test.mjs and node --test test.mjs print different outputs

[ErickWendel]

Version

v20.0.0-pre

Platform

darwin

Subsystem

No response

What steps will reproduce the bug?

No response

How often does it reproduce? Is there a required condition?

npm i -g tap-spec@5.0.0

// index.test.mjs
import { describe, it } from 'node:test'
import { deepEqual } from 'node:assert'
function sum(num1, num2) {
  return num1 + num2
}

describe('Math test suite', () => {
  it('it should sum two integers', () => {
    const expected = 10
    const current = sum(8, 2)
    deepEqual(current, expected)
  })
})

./node --test  watch-mode-tests/index.test.mjs | tap-spec

  Subtest: /Users/erickwendel/Downloads/projetos/node/watch-mode-tests/index.test.mjs

    ✔ /Users/erickwendel/Downloads/projetos/node/watch-mode-tests/index.test.mjs

and

./node watch-mode-tests/index.test.mjs | tap-spec 

  Subtest: Math test suite

        # Subtest: it should sum two integers
        ok 1 - it should sum two integers
          ---
          duration_ms: 0.660125
          ...
    ✔ Math test suite

or in a folder

./node --test watch-mode-tests | tap-spec

  Subtest: /Users/erickwendel/Downloads/projetos/node/watch-mode-tests/index.test.mjs

    ✔ /Users/erickwendel/Downloads/projetos/node/watch-mode-tests/index.test.mjs

What is the expected behavior?

Both should output

  Subtest: Math test suite

        # Subtest: it should sum two integers
        ok 1 - it should sum two integers
          ---
          duration_ms: 0.660125
          ...
    ✔ Math test suite

What do you see instead?

when using --test

  Subtest: /Users/erickwendel/Downloads/projetos/node/watch-mode-tests/index.test.mjs

    ✔ /Users/erickwendel/Downloads/projetos/node/watch-mode-tests/index.test.mjs

Additional information

@nodejs/test_runner @nodejs/tooling

[MoLow]

@ErickWendel this seems to be a duplicate of #43344
can you confirm?

[ErickWendel]

@ErickWendel this seems to be a duplicate of #43344 can you confirm?

I think they're similar but different. In that issue, it's the actual parser that would be modified (and might affect it here also) but here I mentioned that adding the flag --test outputs only the file path that is being executed and without it the complete test results.

I think --test would print the complete result

[MoLow]

@ErickWendel yes that is the same issue

[ErickWendel]

@ErickWendel yes that is the same issue

I didn't get why. Reading that issue means that the output for the test count is incorrect and here are different outputs for running with our without the --test flag.

[cjihrig]

@MoLow is right. The output is different because the CLI runner reports each file as a TAP test point. Currently, the output from each child process is only shared when there is a failure in that file. The TAP parser PR addresses this.

I think we can close this as a dupe.