On Tue, Dec 5, 2017 at 4:37 AM, Scott Marlow <smarlow@redhat.com> wrote:
It would be great if we could have a branch that includes all of the commits that we are considering to merge at a particular time of day, such that we would run the TCK against that branch, only once a day. 

Can this be done that often? I had in my mind that if we did one of these it would amount to stealing one of the regular runs, but perhaps that's not the case.

Now, I don't think we'd want to do these anywhere near that often, but it's good to know what the limits would be. For example, I could imagine doing 10 of these over the course of a WF release, but by luck or whatever 3 of them come in the same week.

+1 to using a branch. We have a branch like that, master-ignore, that we use for batching up PRs to test as a group before merging. I wouldn't want to use master-ignore for this, but a differently named branch run the same way sounds good.
If one of the changes cause a TCK failure, none of them get merged (investigation follows that to determine which change caused the failure(s)), if the test succeeds, we can then merge that batch of changes into WildFly master.

We likely would want to avoid running the testing, on days when we haven't merged any changes to the WF testing branching.

Can the TCK be set up to run based on a check for a change in the sha of the head of a branch? So every day at a fixed time it checks the branch, and does nothing if there is no change. If we want a run, we force push the branch before that time. We have CI jobs that check master-ignore that way, except they poll regularly, not just once a day. That works for those as they aren't so resource intensive that we worry a lot about limiting how often they run.
Would that approach help how we merge PRs on master?

I think it could be helpful earlier in the release cycle before merging big changes, and then perhaps late in the release cycle  if we're worried about possible regressions.


On 12/04/2017 09:33 PM, Stuart Douglas wrote:

On Tue, Dec 5, 2017 at 3:40 AM, Brian Stansberry <brian.stansberry@redhat.com <mailto:brian.stansberry@redhat.com>> wrote:

    Great. :)

    One thing I think we need to do is figure out how to get custom TCK
    runs for PR branches. The TCK is a big part of our test coverage,
    and one way to not "use master as a test bed" is to get a check of a
    branch on the TCK before we merge it.

    I know we've gotten TCK runs of ad-hoc branches before, so by
    "figure out" I mean work out how to make that not overly painful,
    come to some sort of consensus on when it's worthwhile, etc.

I think if we were going to do this it should probably be something reviewers can ask for on specific PR. The TCK uses a *lot* more resources than a standard CI run, so we need to make sure we limit it to cases where it is required.


    On Fri, Dec 1, 2017 at 10:04 AM, Alessio Soldano
    <asoldano@redhat.com <mailto:asoldano@redhat.com>> wrote:

        There you go... PR updated to consume the same api jar now
        released as final.


        On Fri, Dec 1, 2017 at 3:30 PM, David Lloyd
        <david.lloyd@redhat.com <mailto:david.lloyd@redhat.com>> wrote:

            On Thu, Nov 30, 2017 at 5:50 PM, Alessio Soldano
            <asoldano@redhat.com <mailto:asoldano@redhat.com>> wrote:
            > As suggested by Brian, I'd like to draw attention to the discussion on
            > https://github.com/wildfly/wildfly/pull/10604
            <https://github.com/wildfly/wildfly/pull/10604> .
            > The PR is an upgrade of the webservices stack, including JBossWS, Apache
            > CXF, JAXB-RI and JAXB API. In particular, the JAXB upgrade is for EE8 and
            > better JDK 9 compatibility.
            > Now, due to the upgrade of the JAXB API spec jar, the PR is essentially
            > stalled since 20 days; the new spec is released as an alpha (as it's been
            > tested within JBossWS only) and that does not satisfy a rule that requires
            > any artifact being pulled to be Final.
            > We're talking about a spec jar, we could simply re-tag that as Final,
            > chances are we won't need changes any time soon there anyway, but as Tomaz
            > pointed out, in principle that would be dishonest.

            My opinion is that you should go ahead and make a .Final
            tag.  In the
            (unlikely?) event that the spec has to be modified for some
            reason, I
            think you could make a 1.0.1.Final tag and call it a "bug fix".

            The alternative is to simply wait.  I don't think there is
            any middle position.

            > While I see the point in requiring that only sufficiently stable upgrades
            > are applied to the codebase, I'm wondering whether, maybe, we're going a bit
            > too far with the rules. Brian wrote on this topic: "how to determine that
            > something is good enough to go in without using master as a test bed" ?

            I don't think we are; I agree with the policy as it stands.             If you
            look at it in terms of being able to release at any time,
            then it
            follows that everything _must_ be stable.

            - DML

        wildfly-dev mailing list
        wildfly-dev@lists.jboss.org <mailto:wildfly-dev@lists.jboss.org>

    --     Brian Stansberry
    Manager, Senior Principal Software Engineer
    Red Hat

    wildfly-dev mailing list
    wildfly-dev@lists.jboss.org <mailto:wildfly-dev@lists.jboss.org>

wildfly-dev mailing list

Brian Stansberry
Manager, Senior Principal Software Engineer
Red Hat