On Thu, Jan 31, 2019 at 07:08:14PM +0100, Miro Hrončok wrote:
On 31. 01. 19 16:32, Michal Domonkos wrote:
>On Wed, Jan 30, 2019 at 6:46 PM Miro Hrončok <mhroncok(a)redhat.com> wrote:
>>Based on the entire discussion so far, here's my proposal:
>>
>> - we change this to a system wide change
>> - we move it to Fedora 31
>> - we retire the packages from rawhide as soon as f30 is branched regardless
of
>>the dependent packages
>> - packages with broken deps / FTBFS due to this will be retired if not fixed
>>by beta freeze
>>
>>Contingency mechanism:
>>
>> - if some process (releng or similar) needs the packages in order to ship
>>Fedora 31, the packages are added into a designated copr repo maintained by the
>>person/team responsible for the tool that needs yum (or other packages retired)
>>
>> - if the above is not possible and the packages are indeed needed in the
>>actual f31 repos, packages are unretired but the person/team responsible for the
>>tool that needs yum maintains them as long as they need them and retires them
>>once that is no longer true
>
>+1
>
>As an alternative solution, based on a discussion with Neal Gompa
>today on IRC, I propose the following:
>
> - we remove python-urlgrabber from the original change proposal
>(i.e. keeping it in F30)
> - we proceed with the retirement of the rest of the YUM stack in F30
> - we make sure the kojid PR[1] is merged in time for F30
>
>This is based on the following two facts:
>
> - python-urlgrabber seems to be the last component of the YUM stack
>that turns this proposal into a "system-wide" change, due to a number
>of infra bits that require it (sigul, koji-containerbuild, osc or
>imagefactory). Therefore, if we just postpone the removal of
>python-urlgrabber to F31 and merge that kojid PR, we could perhaps
>agree on re-qualifying the change as "self-contained" (plus, there's
>also the possibility of porting[2] python-urlgrabber to Python 3, but
>that's for a separate discussion)
> - the kojid PR[1] is also in-line with another F30 change[3], so
>there should be enough incentive to have it merged
>
>Before I go ahead and edit the proposal: Does this variant make sense to you?
It does to me. But well, I've been called a Python 2 deletionist
before, so not sure if I'm not biased :)
FWIW, I like this "alternative proposal" better too.
Zbyszek