Open Stack

On 27 February 2016 at 00:24, Jeremy Stanley <fungi at yuggoth.org> wrote:
> On 2016-02-26 11:13:52 +0000 (+0000), Neil Jerram wrote:
>> I understand the semantic versioning algorithm for calculating a new
>> version.  But what do I run, in a git repository, to do that calculation
>> for me, and output:
>>
>> - the new semantic version that would be used if I asked for a formal
>> release to PyPI
> [...]
>
> To my knowledge, we have no automation to do this for you. Deciding
> what the next version number of a project should be based on any
> backward incompatibility, new features and so on, is a human-made
> determination. Someone identifies these factors a bit subjectively
> based on the published guidelines and history/release notes and then
> pushes a tag with a new version number based on that determination.

We certainly do -
http://git.openstack.org/cgit/openstack/oslo-specs/plain/specs/juno/pbr-semver.rst
was the spec for it. All it relies on is the use of git tags for
release, + the sem-ver pseudo-header to provoke major and minor
version bumps.

> If you're using PBR in an unreleased repository state, it can
> auto-calculate a temporary "dev" version based on the lowest
> possible next version number which sorts after the most recent tag,
> but that's really only a reflection of what the next version should
> be _if_ all the changes since the last tag are trivial bug fixes.

No - its much more than that. See
http://docs.openstack.org/developer/pbr/#version

-Rob

-- 
Robert Collins <rbtcollins at hpe.com>
Distinguished Technologist
HP Converged Cloud