Having multiple branches to control the version of Keycloak is not the
right approach. IMO Travis should just test with the latest Keycloak
release, then we should have a CI job in Central CI that checks with
Keycloak and RHSSO master.
On 12 June 2017 at 11:18, Bruno Oliveira <bruno(a)abstractj.org> wrote:
Sure. We actually have this issue:
https://issues.jboss.org/
browse/KEYCLOAK-4985.
Which we could catch earlier, if we had the CI running tests against
the changes on Keycloak repo.
So the idea is pretty much it:
- latest: test the latest released version of Node.js adapter, against
the latest stable version of the Keycloak server (we already do this).
- master: test the latest changes from the Node.js adapter, against the
latest changes on Keycloak server.
Another alternative, if you don't like this idea, is to have a build matrix
on Travis. The same idea from keycloak server tests. Or we can do nothing.
The solo porpuse of this change is to guarantee that we catch issues
like this earlier.
On 2017-06-12, Stian Thorgersen wrote:
> This doesn't make sense to me. Can you elaborate a bit more? The reason
it
> makes sense to have two branches for quickstarts is that one branch is
Dev
> and the other is the latest release. For node.js I can't see that being
the
> case as there's "proper" releases and tags and such stuff
>
> On 9 Jun 2017 6:25 pm, "Bruno Oliveira" <bruno(a)abstractj.org>
wrote:
>
> Good morning, I would like to propose the creation of 2 branches for
> Node.js modules following the same approach from the quickstarts:
>
> - latest: stable
> - master: development
>
> The initial motivation behind this, is to enable Travis to test these
> modules
> against the latest change on Keycloak server.
>
> Thoughts?
>
> --
>
> abstractj
> _______________________________________________
> keycloak-dev mailing list
> keycloak-dev(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/keycloak-dev
--
abstractj