> Is actually prefer if there was a way to keep this out of the
code, or if
> they'd adopted Kubernetses format for this instead of inventing thier own.
It is the same format.
Ah, it is the same format some kubernetes projects use (like
https://github.com/kubernetes/test-infra/blob/master/CODEOWNERS). The
main repo uses their own OWNERS which is slightly different.
Martin
On Fri, Sep 8, 2017 at 2:15 PM, Martin Sivak <msivak(a)redhat.com> wrote:
>> Not sure what are they using to identify people, are these GitHub usernames?
>> This feels a bit like lock-in.
>
> @usernames or emails
>
>> It might be better to enforce using email addresse in the file.
>
> Should work according to the article.
>
> Is actually prefer if there was a way to keep this out of the
code, or if
> they'd adopted Kubernetses format for this instead of inventing thier own.
It is the same format.
>
>
> Martin
>
> On Fri, Sep 8, 2017 at 1:39 PM, Barak Korren <bkorren(a)redhat.com> wrote:
>>
>>
>> בתאריך 8 בספט׳ 2017 14:18, "Martin Sivak" <msivak(a)redhat.com>
כתב:
>>
>> Hi,
>>
>> I recently noticed GitHub enabled a feature that allows specifying
>> code owners for different pieces of code:
>>
>>
https://github.com/blog/2392-introducing-code-owners
>>
>> It should supposedly automatically
>>
>> add the proper reviewers to patches.
>>
>>
>> Not sure what are they using to identify people, are these GitHub usernames?
>> This feels a bit like lock-in.
>>
>> It might be better to enforce using email addresse in the file.
>>
>> Is actually prefer if there was a way to keep this out of the code, or if
>> they'd adopted Kubernetses format for this instead of inventing thier own.
>>
>>
>> We have similar feature enabled in Gerrit and it might make sense for
>> our GitHub specific projects to do the same.
>>
>>
>> Sure, why not. Its been very useful in Gerrit.
>>
>> (It might even make sense
>> to follow the same format in Gerrit)
>>
>>
>> If soneone would contribute a parser to the GitHub format (which is to say,
>> something that would scan a commit and yield a list of addresses). We could
>> make it work with a hook or a Jenkins job.