[Pulp-list] Build Numbers

[Jeff Ortel]

On 02/16/2012 08:17 AM, Jay Dobies wrote:
> We've started to build v1 candidates but it's still using the old 0.xyz
> numbering scheme.

Agreed.

That needs to be changed to 1.0-x for the 1.0 release
> and we should probably do it sooner rather than later.

I think it plays like this for v1.

v1/testing contains:
0.266 (QE build)
0.267 (QE build)

READY TO RELEASE

version (pulp_v1) = 1.0.0

BUILD [STABLE]

v1/stable contains:
1.0.0

Time marches on ..... fixes

v1/testing contains:
1.0.1 (QE build)
1.0.2 (QE build)
1.0.3 (QE build)

READY TO RELEASE

version (pulp_v1) = 1.1.0

BUILD [STABLE]

v1/stable contains:
1.1.0

Time marches on ..... fixes

v1/testing contains:
1.1.1 (QE build)
1.1.2 (QE build)
1.1.3 (QE build)

and, on-and-on ...


>
> One added benefit of this is that for people who misconfigure their
> pulp.repo file, the 1.x versions will still take preference over the dev
> builds. Especially with the expected instability of the dev builds
> during the v2 migration, we really need to guide people towards 1.x by
> default and this would be another safeguard against accidentally
> installing dev.

Agreed.

>
> There are two ways we could do it:
> * Branch the v1 stream now and cherry pick over the fixes, then edit the
> spec file in the branch. We'd have to cherry pick over bug fixes, but
> we're almost done with those anyway.

I created the pulp_v1 branch which will serve as /master/ for the v1 
stream.  The pulp_v1 branch is based on pulp-0.0.267-1.  Going forward, 
developers need to either work on pulp_v1 or cherry pick from master as 
appropriate.

> * Edit the spec file in master and then reset it back to 0.* after we
> branch.

I think we should just leave master (dev) as it is.

>
> I don't really care which we take, just that we take one soon since
> we're already at the candidate builds stage.
>
> It's your call on what we do with grinder/gofer. It'd be nice to have
> them look like released versions (grinder is at 0.0.136) but it's a much
> smaller issue than actually marking Pulp as 1.0.
>