Open Stack

On Tue, Nov 29, 2016 at 11:53 AM, Takashi Yamamoto
<yamamoto at midokura.com> wrote:
> release team,
>
> can we (networking-midonet) branch stable/newton from a past commit
> with a RC tag, backport some changes [1], and then cut the first release
> on the branch?

to answer myself, RC or beta-looking tag doesn't seem to be allowed for
release:independent projects. [1]
so i went ahead with 3.0.0. [2]

[1] https://github.com/openstack/releases/blob/745554fdec87b18fe0a39fa25cdc481b23f28d24/openstack_releases/versionutils.py#L48-L49
[2] https://review.openstack.org/#/c/404078/

>
> [1] some addititonal features without db migrations (qos, lbaasv2, ...) and
> removal of some unsupported code (lbaasv1, ...)
>
> On Fri, Nov 25, 2016 at 11:32 PM, Ihar Hrachyshka <ihrachys at redhat.com> wrote:
>>
>>> On 25 Nov 2016, at 15:23, Takashi Yamamoto <yamamoto at midokura.com> wrote:
>>>
>>> On Fri, Nov 25, 2016 at 8:02 PM, Ihar Hrachyshka <ihrachys at redhat.com> wrote:
>>>>
>>>>> On 25 Nov 2016, at 11:02, Takashi Yamamoto <yamamoto at midokura.com> wrote:
>>>>>
>>>>> On Fri, Nov 25, 2016 at 6:54 PM, Ihar Hrachyshka <ihrachys at redhat.com> wrote:
>>>>>>
>>>>>>> On 25 Nov 2016, at 09:25, Takashi Yamamoto <yamamoto at midokura.com> wrote:
>>>>>>>
>>>>>>> On Fri, Nov 25, 2016 at 5:18 PM, Ihar Hrachyshka <ihrachys at redhat.com> wrote:
>>>>>>>>
>>>>>>>>> On 25 Nov 2016, at 05:26, Takashi Yamamoto <yamamoto at midokura.com> wrote:
>>>>>>>>>
>>>>>>>>> hi,
>>>>>>>>>
>>>>>>>>> networking-midonet doesn't have stable/newton branch yet.
>>>>>>>>> newton jobs failures are false alarms.
>>>>>>>>>
>>>>>>>>> branching has been delayed because development of some futures
>>>>>>>>> planned for newton has not been completed yet.
>>>>>>>>>
>>>>>>>>> the plan is to revert ocata-specific changes after branching newton.
>>>>>>>>
>>>>>>>> I don’t think it’s a good idea since you will need to tag a release on branch creation, that is supposed to be compatible with next releases in that same branch.
>>>>>>>
>>>>>>> can't we create the tag after the revert?
>>>>>>>
>>>>>>
>>>>>> No, that’s release team requirement that they branch on a release tag.
>>>>>
>>>>> ok, i didn't know the requirement. thank you.
>>>>>
>>>>>>
>>>>>>> anyway no one think this is a good idea.
>>>>>>> it's just an unfortunate compromise we ended up.
>>>>>>> we are trying to make the schedule better for next release.
>>>>>>
>>>>>> It would make more sense to tag on a compatible commit from the past and consider it a first stable release. (Of course it means that feature development would need to be aligned appropriately.)
>>>>>
>>>>> in that case, can we backport the features?
>>>>> (namely qos and lbaas drivers are in my mind)
>>>>
>>>> No, I don’t think so. Though maybe we can release an RC as the first tag in the branch and backport features before releasing a final version? I dunno, I guess you will need to talk to OpenStack release folks on how to proceed.
>>>
>>> is it a release team matter?
>>> i thought these were a policy inside neutron.
>>> after all networking-midonet is release:independent.
>>
>> Neutron does not override global policies. I explicitly asked during the last summit if we can branch before a tag; the answer was no, it’s not an option.
>>
>> Adding [release] tag since it becomes a matter beyond neutron.
>>
>> Ihar