Open Stack

From: Dave Walker <email at daviey.com<mailto:email at daviey.com>>
Reply-To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Date: Friday, July 22, 2016 at 2:19 PM
To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: Re: [openstack-dev] [kolla] Please start getting in the habit of breaking up containers from ansible changes


On 22 July 2016 at 21:35, Steven Dake (stdake) <stdake at cisco.com<mailto:stdake at cisco.com>> wrote:
>
> Hey folks,
>
> I know it doesn't make a lot of sense to break up containers from ansible changes to people outside the core review team, but for anything with backport potential, please do so.  We are considering in Occata splitting the kolla repo into two (kolla = containers & build, kolla-ansible = playbooks).  I think the timing is right after we branch Kolla Newton, but I don't want to crater our backport process in the process.  By keeping the changes separate we can still have a tidy backport experience.
>
> Even for small changes - 2-3 liner, please break them up using Partial-Bug.
>
> Core reviewers please start enforcing this.
>
> TIA!
> -steve
>

Hi Steve,

Why would this cause a problem in current Master?  As I understand it, you want to make sure that changes that touch both Dockerfiles and Playbooks are in isolated commits so they can be backported.  However, this surely won't be relevant until Newton is cut and Occata is opened, as Newton is remaining as a single tree.  So, the splitting of commits is only relevant in Occata+1, where splitting will already be enforced - by the splitting of the trees in Occata?  I say O+1, as split trees will only start in O.. So for Newton, O commits will still backport cleanly... as they will be separated by the nature of the tree split.

Or... Have I horribly misunderstood your push?

With Occata, will kolla and kolla-ansible have common ancestry? As in, will they both be based on current Master with irrelevant files removed from each tree?

Dave,

This causes no problem in current master or in Newton.

Newton, Mitaka, and Liberty will be one repository.  I want to run a semi-experiment to see if people complain about splitting up the changes to the point that it causes harm to the project and want to get the core reviewer team in the habit for looking for those sorts of issues.  Its a slight change in our workflow.

With Occata and the common ancestry question, if I were doing the work, I'd copy one repo to another, and remove irrelevant files as to not lose history.  What we end up doing will likely be driven by community consensus not my best guess  at speculation of how it should be done so YMMV.

Regards
-steve

Regards
-steve


--
Kind Regards,
Dave Walker
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160722/48097d8c/attachment.html>