Re: [rdo-list] Adding backwards compatible tripleo-heat-templates packaging
On Mon, Sep 5, 2016 at 10:47 AM, Javier Pena <javier.pena(a)redhat.com> wrote:
----- Original Message -----
> On Fri, Sep 2, 2016 at 6:04 PM, Alan Pevec <apevec(a)redhat.com> wrote:
> >> * A single srpm if at all possible
> >
> > Why is that a requirement? Usually in rpm world you create separate
> > parallel installable compat package by adding version to the package
> > name e.g. we had python-webob1.2 in the past.
> > THT mitaka is 2.x so what about creating a new package
> > openstack-tripleo-heat-templates2 ?
>
> We could do that and it's ok, really. It just means a new package
> process for every release of OpenStack. There is no real clean
> mapping historically of THT version to OpenStack version. How does
> someone know that tht2 is mitaka? When we reach ocata and have to
> have a BC package, it's tht5 which is newton. (We could simply name
> it openstack-tripleo-heat-templates-mitaka).
I think something like this (having an openstack-tripleo-heat-templates-compat-mitaka
package or similar) would work best.
This would also allow us to decide when to drop support for older releases. Newton will
have a mitaka-compat package, but we'll sure drop it in a later release, won't we?
We could then just control which packages are built for each release using rdoinfo.
Javier
If we do this, *every* release will drop a package and add a package.
The tripleo team won't commit to supporting more than 1 version back
currently, so -mitaka would only be included with the newton release.
In Ocata, we'd drop mitaka and add -newton. This is part of the
reason that a single tarball made things easier. Just get the right
source tarball for old release and build it in a single package. An
update per release in a single spec file is generally easier than a
new package.
Mike
>
> The proposal currently means that we have to change 1 spec file value
> per release rather than go through a totally new package each release.
>
> Mike
>
>
> >
> > Alan
>
> _______________________________________________
> rdo-list mailing list
> rdo-list(a)redhat.com
> https://www.redhat.com/mailman/listinfo/rdo-list
>
> To unsubscribe: rdo-list-unsubscribe(a)redhat.com
>