[all] Dropping lower constraints testing (WAS: Re: [stable][requirements][neutron] Capping pip in stable branches or not)

Erno Kuvaja ekuvaja at redhat.com
Mon Dec 14 22:09:42 UTC 2020

On Fri, Dec 11, 2020 at 9:10 PM Goutham Pacha Ravi <gouthampravi at gmail.com>

> Hi,
> I hope you won't mind me shifting this discussion to [all] - many projects
> have had to make changes due to the dependency resolver catching some of
> our uncaught lies.
> In manila, i've pushed up three changes to fix the CI on the main,
> stable/victoria and stable/ussuri [1] branches. I used fungi's method of
> installing things and playing whack-a-mole [2] and Brain
> Rosmaita's approach [3] of taking the opportunity to raise the minimum
> required packages for Wallaby. However, this all seems kludgy maintenance -
> and possibly no-one is benefitting from the effort we're putting into this
> as called out.
> Can more distributors and deployment tooling folks comment?
> [1]
> https://review.opendev.org/q/project:openstack/manila+topic:update-requirements
> [2]
> http://lists.openstack.org/pipermail/openstack-discuss/2020-December/019285.html
> [3] https://review.opendev.org/c/openstack/cinder/+/766085
> On Fri, Dec 11, 2020 at 12:51 PM Sorin Sbarnea <ssbarnea at redhat.com>
> wrote:
>> Jeremy nailed it very well.
>> Tripleo already removed lower-constraints from most places (some changes
>> may be still waiting to be gated).
>> Regarding decoupling linting from test-requirements: yes! This was
>> already done by some when conflicts appeared. For old branches I personally
>> do not care much even if maintainers decide to disable linting, their main
>> benefit is on main branches.
>> On Fri, 11 Dec 2020 at 18:14, Radosław Piliszek <
>> radoslaw.piliszek at gmail.com> wrote:
>>> On Fri, Dec 11, 2020 at 5:16 PM Ghanshyam Mann <gmann at ghanshyammann.com>
>>> wrote:
>>> >
>>> > Maintaining it up to date is not so worth compare to the effort it is
>>> taking. I will also suggest to
>>> > remove this.
>>> >
>>> Kolla dropped lower-constraints from all the branches.
>>> -yoctozepto
>>> --
>> --
>> /sorin
Hello all,

While being frustrated to the point I was willing to throw away the
check-requirements job to get around what I thought failed on my efforts to
fix the lower-constraints job (due to me misreading what actually failed in
the check-requirements job) , I think scrapping the lower-constraints job
would be very counterproductive.

We in Glance have been hands full for the past few cycles and assuming
lower-constraints job actually working as intended has led us to neglect
some of our requirements housekeeping quite a bit. If it had not broken
now, we likely would have neglected it for quite a few cycles more. Due to
fixing the said job I had to fix the minimums in our requirements.txt too.
While I'm not sure maintaining the lower-constraints.txt has direct benefit
for many, it actually keeps us honest with our requirements and prevents
stuff breaking down the line. (Expecting that the lower-constraints job
actually works from now on and highlights when we start breaking up on our
dependency chain.) Yes it's a hideous task to get up to date once you have
neglected it for a long time, but I see it as a very valuable tool to
highlight that I should pay more attention to the requirements and what
versions of dependencies we claim to work with.

- jokke
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20201214/a6929093/attachment.html>

More information about the openstack-discuss mailing list