meta: clarify process for breaking changes

[Trott]

Checklist

• documentation is changed or added
• commit message follows commit guidelines

Affected core subsystem(s)

meta

Description of change

Fixes: #7848

[Trott]

/cc @nodejs/ctc

[jasnell]

Lgtm

[yorkie]

Does the "multiple" mean ">= 2 CTC members"?

[Trott]

Yes. More than one.

[ChALkeR]

The main part LGTM, but I have a question about «trivial changes» (commented inline).
/cc @bnoordhuis, perhaps.

[bnoordhuis]

(A priori assumption: small incremental changes > big non-incremental changes.)

I think some pragmatism around the 48/72 hour rule is necessary because it discourages small incremental changes.

It becomes much more attractive to pile up everything in big potpourri pull requests - something I've been guilty of - but that makes code reviews and release management more difficult.

I wouldn't mind dropping the 48 hour rule. If one or two maintainers sign off on a pull request and the CI is green, it should be good to go, right?

[mscdex]

IMHO I'd prefer to keep the 48/72 hour rule. It gives interested parties a chance to chime in on changes that they may feel strongly about. I also recall instances where code changes were landed too quickly (talking non-security issues here) and those changes ended up causing problems of varying degrees...

[evanlucas]

Yea, IIRC we implemented that rule because changes were landing too fast. I'm +1 on keeping the rule, with the exception of doc changes (and maybe changes that have broken CI or something that needs a quick revert?). It's hard, because if we have the rule, and the rule has exceptions, those should be perfectly clear.

[jasnell]

I'm in favor of keeping the 48/72 rule, if for no other reason than it provides time for our global group of contributors to catch up and properly review the changes.

[Trott]

I've removed the change to the "trival changes" description in the paragraph about the 48/72 hour rule. PTAL. So far, we've got LGTMs from @jasnell and @ChALkeR but no one else...

[misterdjules]

LGTM.

[addaleax]

LGTM

[mscdex]

LGTM

[jasnell]

Thinking out loud: it might make sense to introduce new ctc-review and tsc-review labels for issues that (a) currently require review but have not yet been explicitly added to the meeting agenda.

[mhdawson]

LGTM

[Trott]

There seems to be consensus, so last call for objections! I'll land in approximately 12 hours or so unless someone swoops in to say this is terrible or something.

[Trott]

Landed in e3e3588