I don't have a strong opinion on this but, in my experience, having one
reviewer ought to be sufficient assuming they are familiar with the
code-to-be-changed. I don't believe committers should be able review
their code outside of extraordinary circumstances, such as a very high
priority security issue with no available reviewers. Another alternative
is to actually set up a system of "code owners" for different
components, but maintaining such a list (and reducing your pool of
reviewers) can be a bit tiresome.
Regards,
Rob
On 6/14/2022 3:39 PM, Austin Bennett wrote:
Hi Flagon-Dev,
Do we have a Flagon community standard for when OK to merge a PR?
Inviting us to have a discussion, and to get consensus around community
standards for reviewing PRs/when-OK-to-Merge. Apologies if I missed a
prior discussion/doc on this topic.
I assume we ultimately need a committer to review the work of a
non-committer ahead of merge ( since someone with permissions eventually
has to actually accept/merge/commit the code ). In general, is one
reviewer/committer sufficient?
How does this change if the author of the PR is an existing committer? I
have seen some communities require a committee to review things ( no matter
the status of the author), and others require anyone else -- so that's more
of a community determination. For example, could also be a judgement call
as to whether to tag in the expertise of someone having great depth on a
specific component, or just some eyeballs for sanity check.
Hoping we can try to be *somewhat* specific [ where possible ] about the
sort of expected requirements, and our standards. A result of the
discussion likely will be things that can be summarized and used for
updating http://flagon.incubator.apache.org/docs/contributing/
And, naturally there are exceptions ( ex: very large commits, when
functionality significantly altered [ such a case could warrant a FIP /
Flagon-Improvement-Proposal ], etc ). And, exception in the other
direction -- how sufficiently small to be OK merging without another
reviewer [ of any kind ]?
Thanks,
Austin