lib: fixme defer error to next tick

[tflanagan]

Fixes FIXMEs in lib/internal/child_process.js ref's #4642

...
process.nextTick(callback,ex);}else{this.emit('error',ex);// FIXME(bnoordhuis) Defer to next tick.}returnfalse;
...
process.nextTick(callback,ex);}else{this.emit('error',ex);// FIXME(bnoordhuis) Defer to next tick.}}

[jasnell]

Marking this as a possible semver-minor due to the change in the timing. @bnoordhuis .. thoughts on this one?

[cjihrig]

Wouldn't a change in timing be a semver-major change?

[jasnell]

Possibly yes.

[tflanagan]

I've updated this PR to match master

[jasnell]

@bnoordhuis ... thoughts?

[tflanagan]

/poke @bnoordhuis

[Trott]

CI: https://ci.nodejs.org/job/node-test-pull-request/1692/

[thefourtheye]

Related work: #5251

[jasnell]

@bnoordhuis@tflanagan ... ping

[bnoordhuis]

LGTM if tests pass. I'd make it semver-major just to be safe. CI: https://ci.nodejs.org/job/node-test-pull-request/2027/

[MylesBorins]

should we add this to the v7 milestone?

[imyller]

FIXME states that emitting error should be deferred to next tick. setImmediate does this.

process.nextTick and setImmediate have confusing names, since setImmediate will actually defer emitting the error to next tick and process.nextTick will emit error within current tick cycle.

Thus nextTick has lower latency, but does it matter in this case?

[Trott]

LGTM. Since it's semver-major (out of caution, not because this is actually expected to cause huge problems), I'm going to ping @nodejs/ctc to see if anyone else wants to LGTM or wait wait wait wat no no no this.

Also, since it's been dormant a while, might be good to run one last CI. Also a CITGM because: semver-major.

CI: https://ci.nodejs.org/job/node-test-pull-request/4076/
CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/387/

[mscdex]

IMHO I don't think it has to be until the next actual tick, I think the original idea was to allow the user to have time to set up an error event handler before it's emitted. process.nextTick() accomplishes this without further unnecessary delay. It's also what is used throughout the code base whenever similar situations arise.

[imyller]

Agree with @mscdex

LGTM pending new CI and ctc comments

[cjihrig]

LGTM

[santigimeno]

LGTM

[evanlucas]

Running CI one more time: https://ci.nodejs.org/job/node-test-pull-request/4126/

[jasnell]

ping @nodejs/ctc ... did we want to land this?

[fhinkel]

There's two approvals and no rejection, so I'd say...yes?

[gibfahn]

3 ctc approvals (@bnoordhuis@Trott@cjihrig ) and 2 other collaborator ones ( @imyller@santigimeno) so it should be good to land, probably needs another CI and CitGM run though.

[BridgeAR]

Seems like this is not stalled and it only has to land?

[Trott]

CI: https://ci.nodejs.org/job/node-test-pull-request/9587/

[Trott]

Landed in f2b01cb.

Triageathon 2017!