It would only work If you want "All" your PRs to be "approved". => It would
only work If you want "All" your PRs to be "approved" by the GH Actions.



On Tue, Feb 8, 2022 at 4:00 PM Jarek Potiuk <ja...@potiuk.com> wrote:

> Depends on the scheme you choose.
> It would only work If you want "All" your PRs to be "approved".
>
> But it makes little sense to use "approvals" for that because GitHub
> Actions have "Checks" which are handling this case (and which Can be done
> by GitHub Actions).
>
> See the difference here: https://pasteboard.co/RoaA64wv4NvH.png
>
>
>
> J.
>
>
> On Tue, Feb 8, 2022 at 3:42 PM Chesnay Schepler <ches...@apache.org>
> wrote:
>
>> If GA does not approve the PR wouldn't you not want to merge it in the
>> first place?
>> Requiring a second committer in that case doesn't sound like such a bad
>> idea.
>>
>> On 08/02/2022 15:29, Jarek Potiuk wrote:
>> > Yes it is. But you are not able to distinguish commiters vs. GA. So if
>> you
>> > set +2 you will need +2 approval from committers when GA does not
>> approve
>> > it - which is pretty useless if you do not intend to have 2 approvals
>> from
>> > committers.
>> >
>> > On Tue, Feb 8, 2022 at 3:27 PM Chesnay Schepler <ches...@apache.org>
>> wrote:
>> >
>> >> Is it not possible to control the number of approvals that are
>> required?
>> >>
>> >> So if a project has N github actions that verify stuff, then the
>> project
>> >> configures this to N+1.
>> >>
>> >> Admittedly this would mean that every project has use this properly.
>> >>
>> >> On 08/02/2022 15:14, Jarek Potiuk wrote:
>> >>> In short - just to explain why:
>> >>>
>> >>> The "protected branches" feature is the way how to make sure that the
>> >> code
>> >>> is looked at and approved by at least 1 commiter in order to be
>> merged.
>> >>> This is a strong protection - not only from UI but also prevents you
>> from
>> >>> fast-forward the "prtected branch" to the tip of the branch that has
>> not
>> >>> been approved.
>> >>> Currently, you could write a github workflow (like the one that
>> Michael
>> >>> pointed at) where you have no human in the loop to approve such
>> change.
>> >>> GitHub actions run acts as a "commiter" and approval counts as an
>> >> approval
>> >>> from committer. Once approved, merge can also happen using the
>> workflow.
>> >>> This means that even if you have "protected branch" you could have a
>> >>> workflow that allows to merge a code that no human looked at in the
>> >>> "protected branch".
>> >>>
>> >>> The settings mentioned by Gavin means that the workflow can still
>> perform
>> >>> the approval (this wil continue to work) but it will be equivalent to
>> >>> non-committer approval (which does not unblock the PR from merging).
>> >>>
>> >>> J.
>> >>>
>> >>> On Tue, Feb 8, 2022 at 2:40 PM Michael A. Smith <mich...@smith-li.com
>> >
>> >>> wrote:
>> >>>
>> >>>> Not sure if you're asking for in-Apache examples, but GitHub has an
>> >> example
>> >>>> here:
>> >>>>
>> >>>>
>> >>>>
>> >>
>> https://docs.github.com/en/code-security/supply-chain-security/keeping-your-dependencies-updated-automatically/automating-dependabot-with-github-actions?learn=dependency_version_updates&learnProduct=code-security#approve-a-pull-request
>> >>>>
>> >>>> On Tue, Feb 8, 2022 at 07:34 Gary Gregory <garydgreg...@gmail.com>
>> >> wrote:
>> >>>>> Do you have an example of such a GH Action?
>> >>>>>
>> >>>>> Gary
>> >>>>>
>> >>>>> On Tue, Feb 8, 2022, 07:25 Gavin McDonald <gmcdon...@apache.org>
>> >> wrote:
>> >>>>>> Hi All,
>> >>>>>>
>> >>>>>> A recent feature pointed out by Jarek :
>> >>>>>>
>> >>>>>>
>> >>>>>>
>> >>
>> https://github.blog/changelog/2022-01-14-github-actions-prevent-github-actions-from-approving-pull-requests/
>> >>>>>> is to be able to disable GH Actions from approving PRs.
>> >>>>>>
>> >>>>>> Infra intends to untick this box and disable this feature.
>> >>>>>>
>> >>>>>> Any concerns?
>> >>>>>>
>> >>>>>> Anyone uses this feature? If so why.
>> >>>>>>
>> >>>>>> --
>> >>>>>>
>> >>>>>> *Gavin McDonald*
>> >>>>>> Systems Administrator
>> >>>>>> ASF Infrastructure Team
>> >>>>>>
>> >>
>>
>>

Reply via email to