Re: changing merge rules

Jason Yellick <jyellick@...>


We had relaxed on Gerrit to allow self+2 on changes which did not merit additionally scrutiny, and this was a big improvement in productivity from my perspective.  I think it's time we finally removed the second +2 entirely.


----- Original message -----
From: "Yacov" <yacovm@...>
Sent by: fabric@...
To: "David Enyeart" <enyeart@...>
Cc: "Christopher Ferris" <chris.ferris@...>, fabric <fabric@...>
Subject: [EXTERNAL] Re: [Hyperledger Fabric] changing merge rules
Date: Wed, Dec 11, 2019 2:28 AM

I see having a more fine grained access control over who can merge changes to sub-components in the code as a natural extension of this policy change.

From:        "David Enyeart" <enyeart@...>
To:        "Christopher Ferris" <chris.ferris@...>
Cc:        fabric <fabric@...>
Date:        12/11/2019 09:11 AM
Subject:        [EXTERNAL] Re: [Hyperledger Fabric] changing merge rules
Sent by:        fabric@...


Now that we are on GitHub we do have the option to use CODEOWNERS to further specify a set of reviewers with domain knowledge in certain areas of the code, but I think that could come later if we see a need for it in selective areas.
And agree with Yacov about when in doubt get a second review. We could use PR Assignee field to 'nominate' additional reviewers.

Dave Enyeart

"Christopher Ferris" ---12/10/2019 05:25:20 PM---We've brought this up before, but maybe it deserves to be revisited.

From: "Christopher Ferris" <chris.ferris@...>
To: fabric <fabric@...>
Date: 12/10/2019 05:25 PM
Subject: [EXTERNAL] [Hyperledger Fabric] changing merge rules
Sent by: fabric@...

We've brought this up before, but maybe it deserves to be revisited.

Originally, we added the 2+2 because there was a perception in the community that IBMers were being less critical of the contributions by other IBMers.

It served its purpose, but it does raise the bar and especially for obvious things that really don't merit the extra effort.

We have the abilities on GH to prevent self review.

Maybe the time has come as we have moved to GH that we think about lowering the bar and making it less burdensome on the maintainers that are doing reviews by allowing them to merge with a single NACR.

When there's a more substantive change, a maintainer can always seek additional eyes by adding reviewers.

I’d like to propose we go with a single NACR across the Fabric repo-scape.

Please respond on this thread if you concur, or have concerns. If we have a majority, we can ask Ry to make the necessary changes to the merge policy.





Join { to automatically receive all group messages.