This is really great news. Thanks a lot for all the effort Timo, Francesco and everyone else who was involved! I believe that this will make it a lot easier for our users to use any Scala version they want with Flink's vanilla distribution :-)
Cheers, Till On Thu, Dec 30, 2021 at 4:03 PM David Morávek <d...@apache.org> wrote: > Great job! This brings the scala-free effort close to the finish line! > > D. > > On Thu, Dec 30, 2021 at 3:08 PM Timo Walther <twal...@apache.org> wrote: > > > Hi everyone, > > > > The new module flink-table-planner-loader replaces > > flink-table-planner_2.12 and avoids the need for a specific Scala > > version in downstream projects. It is included in the Flink distribution > > under /lib. For backwards compatibility, users can still swap it with > > flink-table-planner_2.12 located in /opt. > > > > As a consequence, flink-table-uber has been split into > > flink-table-api-java-uber, flink-table-planner(-loader), and > > table-runtime. flink-sql-client has no Scala suffix anymore. > > > > It is recommended to let new projects depend on > > flink-table-planner-loader (without Scala suffix) + flink-table-runtime > > in provided scope next to the API. > > > > Note that the distribution does not include the Scala API by default. > > Scala users need to explicitly add a dependency to flink-table-api-scala > > or flink-table-api-scala-bridge. > > > > We added a new README that can be used for future reference about all > > modules that we provide: > > > > https://github.com/apache/flink/blob/master/flink-table/README.md > > > > To the best of our knowledge, currently only the Hive compatibility > > layer needs the old flink-table-planner_2.12. Connectors developers and > > user programs should be fine with flink-table-planner-loader. > > > > For more information see also: > > > > https://issues.apache.org/jira/browse/FLINK-25128 > > > > > > Regards, > > Timo > > >
