MOM5 icon indicating copy to clipboard operation
MOM5 copied to clipboard

MOM5 Governance

Open aidanheerdegen opened this issue 6 years ago • 8 comments

In this PR https://github.com/mom-ocean/MOM5/pull/198 @StephenGriffies raised some governance and other issues which need to be addressed.

--If the pull request was not necessary, then why was it still hanging around?

--I see now there are 8 other pull requests. Which ones are also redundant or problematic? Should the redundant pull requests be removed?

--How often are the pull requests tested for consistency with the main repo? When the pull request says that all tests are fine, then is that consistency rechecked when another pull is made?

I have been thinking that it would be useful to formalise some of this governance stuff. I think it doesn't get dealt with because no-one is quite sure who is responsible, how it should be done etc.

I have some proposals:

  1. Regularish MOM5 technical meetings (maybe just carve out 10 mins after MOM meetings every month?)
  2. Make and document in the Wiki some protocols for dealing with pull requests e.g. assign a PR to someone to deal with -- we should use the tools GitHub provides to help with this, regularly review PR status, no direct commits to master?
  3. Create some GitHub projects for achieving shared goals (improved testing!)

aidanheerdegen avatar Jan 09 '19 23:01 aidanheerdegen

@aidanheerdegen many thanks for bringing these questions to the group. I am happy to be part of the discussions and think your proposals are good.

StephenGriffies avatar Jan 10 '19 01:01 StephenGriffies

Some more ideas for protocols:

  1. Known group of committers, and process for adding new committers
  2. All PRs should be signed off by one other commiter (more eyes = fewer mistakes)

aidanheerdegen avatar Jan 10 '19 04:01 aidanheerdegen

Addressing your other points @StephenGriffies:

How often are the pull requests tested for consistency with the main repo?

Every time there is another commit pushed to the branch from which the PR came. Ricardo's incorrect PR had never been tested because it was done before there was Travis CI enabled.

Regarding the testing, it is ONLY compilation testing, and does not currently test the ACCESS-OM target (but that is something I am working on). You can see what models compilations are tested in this example:

https://travis-ci.org/mom-ocean/MOM5/builds/470823462

Pull requests do not check consistency with other PRs, but they do check against updates to the branch they are a PR to, namely master.

aidanheerdegen avatar Jan 10 '19 04:01 aidanheerdegen

@StephenGriffies We will discuss some of these governance issues at our next Technical Working Group meeting, which is scheduled for next Tuesday (15 Jan) 10am AEST. You're welcome to attend, or we can update you on what we discuss.

marshallward avatar Jan 10 '19 23:01 marshallward

Thanks @aidanheerdegen for putting these issues on the agenda. I will be participating in a telecon that will keep me from joining your meeting. So please just update me when you can. Thanks.

StephenGriffies avatar Jan 10 '19 23:01 StephenGriffies

oops. I meant to also thank @marshallward ...

StephenGriffies avatar Jan 10 '19 23:01 StephenGriffies

@StephenGriffies I've shared with you a preliminary document we'll use to flesh out our ideas. Early days, but feel free to take a look and contribute. Will definitely update it after the meeting and get your feedback. If necessary we can have another dedicated meeting to hash out any details.

aidanheerdegen avatar Jan 11 '19 00:01 aidanheerdegen

Thanks much @aidanheerdegen for this material. I will try to look once you have the meeting.

StephenGriffies avatar Jan 11 '19 03:01 StephenGriffies