[Opendnssec-develop] Move to git

[Sara Dickinson]

On 11 Feb 2014, at 07:24, Jerry Lundström <jerry at opendnssec.org> wrote:

> On Mon, Feb 10, 2014 at 4:09 PM, Sara Dickinson <sara at sinodun.com> wrote:
> 
> I have added a few extra bullet points and heading to the workflow page - please review my changes and make sure they all make sense!
> 
> "however remember that this will
mean
> they are dependant on each other for being accepted into the upstream repository."
> ..that this will they are.. ? :)

Fixed. 

>  
> Question: Right now who has permission to run what bot commands?
> 
> Should be all of us.

Excellent!

>  
> In the beginning I think I hear that the plan is that Jerry/Jakob will run every test on every pull request (not the developer)? Is this right? Will you do this immediately on receiving the request or only after review? How long does this take for the production releases - 2 hours I think for both DBs?
> 
> Currently:
> - After review.
> - Only smoke.
> - Only SQLite.
> - 10-20mins.

Thank you for clarifying! 

But - I’m confused. Your previous emails stated:

On 23 Jan 2014, at 06:58, Jerry Lundström <jerry at opendnssec.org> wrote:
> build - Build and run all tests
> test - Build and run only the modified tests, if there are no modified tests then run all tests

And "The initial plan is to run all test because it was the easiest to get working at first (see “build” above).”

But this contradicts what you wrote above as you list only a subset of tests….

1) Is it currently possible to run the mysql or daily tests via a bot command?
2) If we follow the above, then it is still possible, in the beginning that code goes into the repo that breaks the MySQL or daily tests - correct? And if that happens developers need to fix it via another pull request?

I’m not saying this is the wrong approach to start with - I am just trying to get a clear picture of what you and Jakob have decided and communicate this to the team. 

> 
> I can't answer for people but we have the option do allow developers to build and test their feature branch. Doing this does not really remove the requirement to have a separate set of eyes to look at the code.

My point is that IMHO the role of the release manager is not to personally do a formal code review (requirements/design/implementation) of each pull request - that should be left to the developers - rather it is make sure the right people have reviewed it,  do a sanity check of the code and do the mechanics of running the tests/merge. Do people agree with this?

Sara.