On Wed, May 11, 2022 at 1:10 AM Enrico Olivelli <eolive...@gmail.com> wrote: > I am sorry, I missed this discussion. > But until we cut a release we are in time to change our minds, if we > find out that we can do better.
Yes, but the precise point of having a PIP process is to have discussions and formalize decisions. > > > > > It is not that the PR came out of the blue. Obviously every decision > > can be re-visited if there are additional details, though it would be > > better if we get the feedback at the time the proposal. > > > > To reiterate the rationale for going directly to 17: > > > > 1. Requiring Java 11 won't buy us anything new and will at the same > > time require changes from the part of the users. > > 2. 17 is a Java LTS release that will be out for 1 year from the > > moment in which we release Pulsar 2.11 > > 3. It is a stable release with widely available packages for every > > platform and from every Java vendor. > > 4. We are setting up for 4 years of active support of Java 17, > > compared to just 1 year of Java 11 > > 5. There are several source-level features introduced in 12+ that we > > can take advantage of in our codebase > > I understand your points, and I would be really excited to start using > Records and other features (and Valhalla, Loop and Panama as soon as > they are available) > > But on the other side now in the Pulsar ecosystem we have big > enterprises that are not keen on changing JDK so quickly. > > Up to version 2.10 Pulsar still worked well on JDK8. > > We cannot require users to switch from JDK8 to JDK17 while upgrading Pulsar. Why not? We can ask to upgrade to Java 11 but not 17? > > We have been running, building and testing Pulsar on JDK11 for many > major releases (from 2.7 onwards) (and the docker images in 2.10 are > with JDK11) > so it is time to require JDK11. Requiring Java 11 would be pointless as there are very few Java source level features we can take advantage of. > > I believe that the best plan, in the interest of our community and of > the enterprises who choose to switch to Pulsar, > is to still allow all Pulsar components to run on JDK11 (and the > client on JDK8) for 2.11. > > We can switch to requiring JDK17 in 2.12. I do not agree with this assessment and I think the best plan is to continue with the current decision of switching to Java 17.
