async-hooks: internalize setInitTriggerId

[refack]

• replace setInitTriggerId with triggerIdScopeSync

Checklist

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

Affected core subsystem(s)

async_hooks

[refack]

/cc @nodejs/async_hooks

I'm still getting this wrong with 605de08, it does not detect the retrive & reset

not ok 23 async-hooks/test-graph.pipeconnect
  ---
  duration_ms: 0.511
  severity: fail
  stack: |-
    'pipeconnect:1' expected to be triggered by 'pipe:2', but was triggered by 'pipe:1' instead.
    assert.js:48
      throw new errors.AssertionError({
      ^
    
    AssertionError [ERR_ASSERTION]: 1 === 0
        at verifyGraph (/data/iojs/build/workspace/node-test-commit-linuxone/nodes/rhel72-s390x/test/async-hooks/verify-graph.js:89:10)
        at process.onexit (/data/iojs/build/workspace/node-test-commit-linuxone/nodes/rhel72-s390x/test/async-hooks/test-graph.pipeconnect.js:29:3)
        at emitOne (events.js:120:20)
        at process.emit (events.js:210:7)
  ...
not ok 25 async-hooks/test-graph.shutdown
  ---
  duration_ms: 0.510
  severity: fail
  stack: |-
    'getaddrinforeq:1' expected to be triggered by 'tcp:2', but was triggered by 'tcp:1' instead.
    assert.js:48
      throw new errors.AssertionError({
      ^
    
    AssertionError [ERR_ASSERTION]: 1 === 0
        at verifyGraph (/data/iojs/build/workspace/node-test-commit-linuxone/nodes/rhel72-s390x/test/async-hooks/verify-graph.js:89:10)
        at process.onexit (/data/iojs/build/workspace/node-test-commit-linuxone/nodes/rhel72-s390x/test/async-hooks/test-graph.shutdown.js:35:3)
        at emitOne (events.js:120:20)
        at process.emit (events.js:210:7)
  ...
not ok 30 async-hooks/test-graph.tls-write
  ---
  duration_ms: 0.512
  severity: fail
  stack: |-
    'getaddrinforeq:1' expected to be triggered by 'tls:1', but was triggered by 'tcp:1' instead.
    assert.js:48
      throw new errors.AssertionError({
      ^
    
    AssertionError [ERR_ASSERTION]: 1 === 0
        at verifyGraph (/data/iojs/build/workspace/node-test-commit-linuxone/nodes/rhel72-s390x/test/async-hooks/verify-graph.js:89:10)
        at process.onexit (/data/iojs/build/workspace/node-test-commit-linuxone/nodes/rhel72-s390x/test/async-hooks/test-graph.tls-write.js:56:3)
        at emitOne (events.js:120:20)
        at process.emit (events.js:210:7)
  ...

[AndreasMadsen]

I think we should wait with actual code until after we have finished all three discussions outlined in #14238. After that, it will be easier for us to decide on the correct layer of separation.

edit: I would also like the code separation/depreciation to happen separately from things such as the triggerIdScopeSync refactor. Otherwise, I think the PRs will become too big.

[refack]

This PR is my code-and-learn... I personally think better through code... I won't mind if this PR will be closed without landing.

So I agree with your second point, re separation/depreciation/refactor

[AndreasMadsen]

@refack I have created a private PR (https://github.com/AndreasMadsen/node/pull/1) that contains the separation and depreciation I have in mind. I will be happy to discuss refactors such as triggerIdScopeSync from that point of view.

---

I intend to submit a PR similar to https://github.com/AndreasMadsen/node/pull/1 if all three discussion topics as outlined in #14238 ends with us agreeing to deprecate the undocumented API. I'm not entirely convinced we will agree on that and I want to respect the decision process, this is why I'm so careful about discussing too many things at once.

[AndreasMadsen]

Just to avoid confusion, I'm -1 with the current approach but I support the overall idea.

[refack]

Just to avoid confusion, I'm -1 with the current approach but I support the overall idea.

👍

[BridgeAR]

Where do we stand here? Is this something you still want to follow up upon @refack? Due to the -1 I am not sure how much progress is possible here in general at the moment.

[AndreasMadsen]

It is now clear to me that this exact implementation is not the solution. #14238 is a long discussion on some alternatives.

[refack]

(this was a test balloon anyway)