test: add coverage for child_process bounds check

[Trott]

Make sure that monkey-patching process.execArgv doesn't cause
child_process to incorrectly munge execArgv in fork().

This basically is adding coverage for an index > 0 check (see comment
in this commit). Previously, that condition was never false in any of
the tests.

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• commit message follows commit guidelines

Affected core subsystem(s)

test child_process

[cjihrig]

LGTM. I think the link and ESLint comments could be dropped though.

[Trott]

Link and ESLint comments removed.

CI: https://ci.nodejs.org/job/node-test-pull-request/6810/

[Trott]

CI again with adjustment for Windows CLI: https://ci.nodejs.org/job/node-test-pull-request/6815/

[Trott]

Test still failing on Windows. Marking in progress until that gets sorted...

[Trott]

Sample failure on Windows: https://ci.nodejs.org/job/node-test-binary-windows/7064/RUN_SUBSET=1,VS_VERSION=vs2015,label=win2012r2/console

not ok 31 parallel/test-cli-eval
  ---
  duration_ms: 3.517
  severity: fail
  stack: |-
    
    assert.js:81
      throw new assert.AssertionError({
      ^
    AssertionError: '' === '42\n'
        at child.exec.common.mustCall (c:\workspace\node-test-binary-windows\RUN_SUBSET\1\VS_VERSION\vs2015\label\win2012r2\test\parallel\test-cli-eval.js:154:14)
        at c:\workspace\node-test-binary-windows\RUN_SUBSET\1\VS_VERSION\vs2015\label\win2012r2\test\common.js:473:15
        at ChildProcess.exithandler (child_process.js:238:7)
        at emitTwo (events.js:127:13)
        at ChildProcess.emit (events.js:215:7)
        at maybeClose (internal/child_process.js:893:16)
        at Socket.<anonymous> (internal/child_process.js:335:11)
        at emitOne (events.js:117:13)
        at Socket.emit (events.js:212:7)
        at Pipe._handle.close [as _onclose] (net.js:534:12)

[Trott]

So, Windows is not getting the output from console.log(42) but everything else is...

Perhaps related to nodejs/node-v0.x-archive#3871?

Maybe this is some "this OS blocks stdio but that one doesn't" thing? Maybe I'm reaching here...

Relevant code:

child.exec(
  `${nodejs} -e "process.execArgv = ['-e', 'console.log(42)', 'thirdArg'];
                 require('child_process').fork('${emptyFile}')"`,
  common.mustCall((err, stdout, stderr) => {
    assert.ifError(err);
    assert.strictEqual(stdout, '42\n');  // <- Raises AssertionError on Windows Only
    assert.strictEqual(stderr, '');
  }));

/cc @nodejs/platform-windows

[Trott]

Trying to squash everything into a single line to see if that makes a difference (which makes sense if so). New CI: https://ci.nodejs.org/job/node-test-pull-request/6909/

[Trott]

OMG, that did it! Landing soon...

[Trott]

Landed in df69d95

[MylesBorins]

not landing cleanly on v6.x feel free to backport

edit: nvm got it