Open Stack

On 16 October 2015 at 15:14, Fox, Kevin M <Kevin.Fox at pnnl.gov> wrote:

> Then how about the alternate approach of:
>
> For bug fixes, copying an existing in tree unit test and tweaking it
> slightly to test for the bug is sufficient to get it accepted instead of
> forcing it to be rewritten in whatever the new hotness for testing at the
> time thing is, that no one has good examples for, nor has fully agreed on
> what a good example of that looks like, leading to long delays in the big
> fix being accepted? IMHO, bug fixes aren't the place to trail blaze unit
> tests.
>

> The unit tests can be rewritten to be "better" in a follow up review that
> doesn't necessarily need to be back ported.
>

I am pretty confident that the experience you went through was an exception
rather than the rule. We're pretty flexible and allow a patch to go in
(master) with a promise of a follow up, if it's critical backport. The
important thing is to identify that...and this is what Ihar is talking
about in this thread.

It is noteworthy that 'trail blazing' as you say makes the patch difficult
to backport, so any advice given in this direction was clearly pointing you
the wrong way.


>
> Would that work?
>

> Thanks,
> Kevin
>
> ------------------------------
> *From:* Kevin Benton [blak111 at gmail.com]
> *Sent:* Friday, October 16, 2015 2:32 PM
>
> *To:* OpenStack Development Mailing List (not for usage questions)
> *Subject:* Re: [openstack-dev] [neutron][stable] proactive backporting
>
> We can't put in code and just hope for testing later. The tests are even
> more important in back-ports because there could be unexpected differences
> in the stable branch that make the patch not work correctly.
>
> However, we do need to make sure that people aren't complaining about the
> testing methodology in the back-ports because it's too late for that.
>
> On Fri, Oct 16, 2015 at 2:23 PM, Fox, Kevin M <Kevin.Fox at pnnl.gov> wrote:
>
>> It would also help if the process could split out bug fixes from unit
>> tests. I had a bug fix get stuck while the unit tests were bikesheded for a
>> while, and then the delay of not getting quickly backported to the stable
>> branches then kicked in. All the while I had to patch production clouds by
>> hand with a non upstreamed fix. I'm all for having unit tests to ensure the
>> bugs don't creep back in, but regression bugs should be fixed as quickly as
>> possible.
>>
>> Thanks,
>> Kevin
>> ________________________________________
>> From: Edgar Magana [edgar.magana at workday.com]
>> Sent: Friday, October 16, 2015 2:04 PM
>> To: OpenStack Development Mailing List (not for usage questions)
>> Subject: Re: [openstack-dev] [neutron][stable] proactive backporting
>>
>> + 2 and total support for it.
>>
>> Looking forward to reviewing the first set of them.
>>
>> Edgar
>>
>>
>>
>> On 10/16/15, 5:33 AM, "Ihar Hrachyshka" <ihrachys at redhat.com> wrote:
>>
>> >Hi all,
>> >
>> >I’d like to introduce a new initiative around stable branches for
>> neutron official projects (neutron, neutron-*aas, python-neutronclient)
>> that is intended to straighten our backporting process and make us more
>> proactive in fixing bugs in stable branches. ‘Proactive' meaning: don’t
>> wait until a known bug hits a user that consumes stable branches, but
>> backport fixes in advance quickly after they hit master.
>> >
>> >The idea is simple: every Fri I walk thru the new commits merged into
>> master since last check; produce lists of bugs that are mentioned in
>> Related-Bug/Closes-Bug; paste them into:
>> >
>> >https://etherpad.openstack.org/p/stable-bug-candidates-from-master
>> >
>> >Then I click thru the bug report links to determine whether it’s worth a
>> backport and briefly classify them. If I have cycles, I also request
>> backports where it’s easy (== a mere 'Cherry-Pick to' button click).
>> >
>> >After that, those interested in maintaining neutron stable branches can
>> take those bugs one by one and handle them, which means: checking where it
>> really applies for backport; creating backport reviews (solving conflicts,
>> making tests pass). After it’s up for review for all branches affected and
>> applicable, the bug is removed from the list.
>> >
>> >I started on that path two weeks ago, doing initial swipe thru all
>> commits starting from stable/liberty spin off. If enough participants join
>> the process, we may think of going back into git history to backport
>> interesting fixes from stable/liberty into stable/kilo.
>> >
>> >Don’t hesitate to ask about details of the process, and happy
>> backporting,
>> >
>> >Ihar
>> __________________________________________________________________________
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe:
>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>> __________________________________________________________________________
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe:
>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>
>
>
> --
> Kevin Benton
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20151016/a14830ab/attachment.html>