openstack team mailing list archive
-
openstack team
-
Mailing list archive
-
Message #14080
Re: PyPI uploads for client libs live
On Tue, Jul 3, 2012 at 9:54 AM, Monty Taylor <mordred@xxxxxxxxxxxx> wrote:
>
>
> On 07/03/2012 08:47 AM, Doug Hellmann wrote:
> >
> > On Jul 3, 2012, at 5:57 AM, Thierry Carrez <thierry@xxxxxxxxxxxxx>
> > wrote:
> >
> >> Monty Taylor wrote:
> >>> At the moment, the only people with permission to upload tags is
> >>> the openstack-release team. However, since we're letting client
> >>> libs manage their own versions, I kinda think we should give PTLs
> >>> the right on their own project - so Vish would get tag access to
> >>> python-novaclient, Brian to python-glanceclient, etc.
> >>
> >> Ideally it would be a two-side approval process (PTL +
> >> openstack-release), because openstack-release shouldn't be able to
> >> tag without PTL approval, and openstack-release should still be
> >> kept in the loop before a tag is pushed by PTLs (there are multiple
> >> reasons why a few hours delay before tagging would be a good idea,
> >> and the openstack-release people actually keep track of those).
> >>
> >> That said, we don't have that approval mechanism available yet
> >> (same issue with the core projects release tagging) so in the mean
> >> time we should probably let the small set of individuals with an
> >> understanding of the issues (PTLs + openstack-release) have the
> >> power to do it. Within that group, we can have a soft two-side
> >> approval process (based on IRC pings) to check "everything is OK"
> >> before triggering a release.
> >
> > Could we use gerrit's 2-step approval like we do for other projects
> > and combine it with a fancy tag detector like was just added for
> > DocImpact?
>
> I have an outstanding bug/feature request for gerrit to allow for the
> review of and approval or disapproval of tags.
>
Ah, I actually meant the "magical text" feature you mention below, but
approving tags sounds like a good thing to have.
>
> That being said - if we wanted to go the route you're talking about in
> the mean time, instead of actually using the git tag route we could have
> an additional commit with a magical text in the commit message which, on
> merging, would cause a tag to be created... We've got guys on the team
> who hack gerrit now though - lemme get some feedback on how much it
> would take to actually get proper tag reviewing.
>
Sounds good.
Doug
References