Open Stack

Excerpts from Sean Dague's message of 2015-07-29 09:54:08 -0400:
> On 07/29/2015 09:49 AM, Jeremy Stanley wrote:
> > On 2015-07-29 09:39:46 -0400 (-0400), Sean Dague wrote:
> >> On 07/29/2015 09:29 AM, Matt Riedemann wrote:
> >>> On 7/29/2015 8:17 AM, Matt Riedemann wrote:
> > [...]
> >>>> ValueError: git history requires a target version of
> >>>> pbr.version.SemanticVersion(2015.1.2), but target version is
> >>>> pbr.version.SemanticVersion(2015.1.1)
> > [...]
> >> So, after every release a giant amount of patches all have to land lock
> >> step or everything is broken?
> >>
> >> That seems pretty fragile. Can we revisit whatever decision caused this
> >> issue?
> > 
> > This came with the semver implementation in PBR 0.11, specifically
> > the idea that if your most recent tag is higher than the version you
> > claim to be working toward in setup.cfg then something is terribly
> > wrong and PBR should throw its hands up in the air until you fix
> > your config. Useful in theory, but racy in practice (especially when
> > you have one team of people pushing tags and a different team
> > approving updates to the repo's setup.cfg file).
> > 
> > A suitable compromise might be to add a knob to PBR (probably via a
> > directive in setup.cfg) to emit a warning and fall back on version
> > guessing as if the version entry were not present at all.
> 
> Right, because otherwise we need to account for 1 to 2 days of gate
> blockage after every stable release, which seems broken by design.

Another solution is to move all projects to post-versioning and
stop setting a version in setup.cfg. That would mean having all
projects following the release:cycle-with-intermediary model, and
I intend to propose making that change during mitaka.

Doug