As Jan has checked, the artifacts on full are indeed delivered as part of the release (they were there on 10.0.0) I made several mistakes related to this topic. First, I did not check if the artifacts were there on the apache repo (the user that reported the issue sent me a central repo link that I did not check thoroughly since I trusted the guy), so, assuming they were not there, I filed a PR to make the modules part of the regular build that should have never existed. Third, I make a mistake while implementing the PR. The good news is that there is nothing to do differently than in 10.0.0 regarding python
On Tue, Apr 1, 2025 at 4:27 PM Alex Porcelli <a...@porcelli.me> wrote: > Francisco, > > This is a great time to make all necessary adjustments in the > pipelines [1][2] and build image[3], as we'll need to restart our > 10.1.0 release (due a failure of branching pipeline). Early next week > I expect to use the weekly to update KIE Tools and once merged, moving > forward (and back to) the 10.1.0 release. > > [1] - https://github.com/apache/incubator-kie-kogito-pipelines/tree/main > [2] - > https://github.com/apache/incubator-kie-kogito-runtimes/tree/main/.ci/jenkins > [3] - > https://github.com/apache/incubator-kie-kogito-pipelines/blob/main/apache-nodes/Dockerfile.kogito-ci-build > > > On Tue, Apr 1, 2025 at 7:19 AM Francisco Javier Tirado Sarti > <ftira...@redhat.com.invalid> wrote: > > > > They are regular java artifacts. > > The only caveat is that in order for the Python IT test to work, a Python > > interpreter should be installed in the image doing the test. Another > caveat > > is that some of those IT tests depend on numpy, and numpy breaks > backward > > compatibility between python releases generously (we are not the only > ones > > ;)), so there might be test failures depending on the python and numpy > > versions installed on the testing machine. > > > > On Tue, Apr 1, 2025 at 1:12 PM Alex Porcelli <porce...@apache.org> > wrote: > > > > > Francisco, > > > > > > Sure we can and actually should release all artifacts. > > > > > > However we need someone to review the pipeline and provide additional > > > instructions on how to release and publish those artifacts. > > > > > > - > > > Alex > > > > > > > > > On Tue, Apr 1, 2025 at 6:39 AM Francisco Javier Tirado Sarti > > > <ftira...@redhat.com.invalid> wrote: > > > > > > > Reverted. > > > > If I recall correctly, the reason I did the change (incompletely, > > > because I > > > > forgot to include the runtime module, but this was not possible to > detect > > > > when running locally using snapshots, it will be certainly apparent > when > > > > releasing) was because I see that the artifacts in full profile (but > not > > > in > > > > regular one) are apparently not released and there are users that > would > > > > like to have the python functionality, but it was only available as > > > > snapshot. > > > > So, my question is, can we also release artifacts that are only in > full > > > > profile? > > > > > > > > On Tue, Apr 1, 2025 at 12:10 PM Francisco Javier Tirado Sarti < > > > > ftira...@redhat.com> wrote: > > > > > > > > > Ill fix right away now > > > > > > > > > > On Tue, Apr 1, 2025 at 10:16 AM Jan Šťastný <jstastn...@apache.org > > > > > > wrote: > > > > > > > > > >> Obviously my thoughts collided in "I think kogito-runtimes build > is > > > not > > > > >> broken after:" - I meant the kogito-runtimes build IS broken :-) > > > > >> > > > > >> On Tue, 1 Apr 2025 at 10:03, Jan Šťastný <jstastn...@apache.org> > > > wrote: > > > > >> > > > > >> > Hello all, > > > > >> > > > > > >> > I think kogito-runtimes build is not broken after: > > > > >> > > https://github.com/apache/incubator-kie-kogito-runtimes/pull/3863 > > > > >> > > > > > >> > The change in the linked PR moved the module > quarkus/addons/python > > > > into > > > > >> > the default reactor, instead of the original 'full' profile. > > > > >> > Problem is that this module has dependencies on other > python-related > > > > >> > modules inside kogito-runtimes which are NOT in the default > reactor, > > > > but > > > > >> > still in 'full' profile. > > > > >> > > > > > >> > As a result the build is broken for users not running the 'full' > > > > profile > > > > >> > explicitly - in which case a stale snapshot of the dependencies > > > which > > > > >> are > > > > >> > by default not part of the reactor is used. > > > > >> > > > > > >> > I've commented on the original PR too. > > > > >> > > > > > >> > Similar changes need to be well tested in affected build > "modes", if > > > > >> we're > > > > >> > moving things from profile into default build, we need to test > both > > > > >> builds > > > > >> > - and keep in mind, CI does NOT necessarily run builds in all > the > > > > >> possible > > > > >> > setups (I think it runs full profile, so the problem was not > > > > apparent). > > > > >> > > > > > >> > @Francisco if you could either revert your change or fix the > setup > > > in > > > > a > > > > >> > reasonably short time. > > > > >> > > > > > >> > Regards > > > > >> > Jan > > > > >> > > > > > >> > > > > > > > > > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@kie.apache.org > For additional commands, e-mail: dev-h...@kie.apache.org > >
