Hi all, As mentioned during last Friday's meeting, I believe that we should move any forward upstream development to Quarkus 3 and Jakarta namespace. That's because the effort of developing against both Quarkus 2.x and 3.x will be hardly bearable and the current setup that automatically migrates our upstream branches (based on Quarkus2/JavaEE) to Quarkus 3 through openrewrite is demonstrating all its limitations. Openrewrite is an amazing tool, but it is mostly intended for one-shoot migration so in my opinion our usage pattern is a sort of abuse, and even worse it doesn't allow us to work effectively on proper Quarkus 3 integration.
Related to this we should also discuss which extensions we want to make available with Quarkus 3. Last week I implemented and merged a commit moving the last rules related feature that was still available only through Kogito (the automatic rest endpoints generation from rule units queries) back into Drools. At this point I think that having a Kogito extension for rules that simply replicates what will be available with the Drools one will be unnecessary and only a source of confusion. To recap my proposal is the following: 1. From the main branches create branches 1.x for Kogito and 8.x for Drools. We will consider these branches in pure maintenance mode and backport there only (critical) fixes. 2. Apply the openrewrite script one last time to definitively migrate our main branches to Quarkus3/Jakarta and eventually fix manually any outstanding incompatibility. 3. Remove the Kogito rule extension, check if all the remaining extensions still make sense (we should avoid features duplication and overlapping in different extensions). 4. Implement and publish asap the Quarkus 3 extensions. As a final note on version numbers, I guess that we don't have many alternatives other than tagging the releases made from these new main branches by continuing with 2.45.0 for Kogito and 9.45.0 for Drools. This could be a bit unusual and confusing for users migrating to these newer versions, but the only other possibility that I see would be to jump to 3.0.0 and 10.0.0 which will be probably even worse. Any feedback is welcome. Thanks, Mario
