asymmetric via NixOS Discourse email@example.com writes:
I respectfully disagree.
It seems to me a commit like the one you posted is squarely a case of
something that should go through a PR. I could imagine a similar
commit straight to master introducing a security vulnerability by
mistake, with the only safeguard against that being one person’s
attention, and I think that’s hardly enough.
I think you might misunderstand my point.
I was not the author of that PR, but the merger. I had to squash it on
the command line, because it needed more than just a squash into a
single commit, which is all GitHub supports.
If the submitter had checked the appropriate checkbox, I could have
pushed the changes to the PR and then merged, but I don’t see that
that’s any difference to committing directly. Am I supposed to open
another PR that’s just a squashed version of the other one, and then
potentially wait for yet another person to come along and review that?
OK, but what should happen then? Someone goes and reviews all
already-pushed commits to master looking for problems? Then it seems
better to have the check beforehand, at least it’s mandatory, and we
avoid having problematic commits in master in the first place.
I understand the frustration with having to wait potentially weeks (!)
for a PR to get merged, but this is a general problem we all have - it
seems slightly unfair for committers to have preferential access in
this case. I can see how exceptions might be necessary (e.g. for
security hotfixes), but version bumps like the one linked above IMO
don’t fit the bill.
Committers already have preferential access. That’s what being a
committer is. I can merge my own pull requests, and if you’re proposing
changing that, then I think you’re going to have a lot of very unhappy