Revert of nodejs/node#55623

[Flarna]

The PR #55623 caused a new issue: #56002.

#55623 landed already in 23.2.0 and to my understanding of contributor docs here a revert of a change already part of the releaes requires a TSC decission/discussion.

There is some history on above PR:

• Triggered by require('fs').openSync('\\\\.\\PhysicalDrive2', fs.constants.O_RDWR) is broken #54025 - which is fixed by this PR to my understanding.
• There was already a fix before (path: fix bugs and inconsistencies #54224) (marked as semver major) causeing npm pack crash in Node 23 #55410 and was reverted via Revert "path: fix bugs and inconsistencies" #55414

I don't have the needed windows know how to actually fix the problem on short notice.

Which direction should we go? Revert or wait a bit for a fixup?

[Flarna]

PR to fix the issue instead of reverting: #56110
PR to revert: #56088

[Flarna]

@nodejs/tsc Any opinion how to progress in this?
I assume the contributor docs telling that TSC decide on revert or semver major labels after the fact has a reason.

[targos]

I don't think this change is affected by https://github.com/nodejs/node/blob/main/doc/contributing/collaborator-guide.md?rgh-link-date=2024-11-30T19%3A01%3A50Z#unintended-breaking-changes.
As far as I understand, the change introduced a bug, not a breaking change in the semver sense.

[Flarna]

Well, the result of e.g. path.resolve() differs before and after. Even after the proposed fix it differs.

In general I'm fine with going ahead and merge the fix and let it go it's path into 23.

I assume that implies also that the dont-land- labels should be removed once that happens to let the initial fix float into older release lines.

[aduh95]

Should we transfer this to nodejs/node? I don't think it's up to the TSC as long as there's consensus among collaborators.

[Flarna]

It's more about semver major yes/no.
if yes how to deal with it (add label/revert).
if no we can close this.

[joyeecheung]

It looks like a bug, I also agree that this should be in nodejs/node. You can send a PR to revert and maybe find a releaser who's willing to volunteer to do a patch release, unless someone stops the revert and a consensus cannot be reached, the ball isn't really in TSC's court.

[Flarna]

ok, in that case I will go ahead, approve the fix PR and let it flow into release. I see no need for a revert.