[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user mlipkovich commented on the issue: https://github.com/apache/flink/pull/4655 Thanks! Closing it ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user aljoscha commented on the issue: https://github.com/apache/flink/pull/4655 Perfect! Thanks for working on this. ð I merged, could you please close this PR? ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user mlipkovich commented on the issue: https://github.com/apache/flink/pull/4655 Hi @aljoscha It turned out that exclude filter works if we escape dollar signs. I commited these changes, but I believe it still should be fixed by japicmp What do you think about merging this PR? ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user aljoscha commented on the issue: https://github.com/apache/flink/pull/4655 Thanks! This doesn't help us with our situation, though. ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user mlipkovich commented on the issue: https://github.com/apache/flink/pull/4655 Thansk, Aljoscha Probably I've forgotten to run `clean`. The method `iterate$default$3` is a method which automatically created by Scala for calculation of default value for parameter `keepPartitioning`. I tried different ways to exclude it but it didn't help. Anyway it should be somehow tracked by japicmp so I created the issue there https://github.com/siom79/japicmp/issues/176 ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user aljoscha commented on the issue: https://github.com/apache/flink/pull/4655 A local build of `mvn clean verify` fails for me because the japicmp plugin is complaining. The relevant section of the diff output from japicmp is ``` ***! MODIFIED CLASS: PUBLIC org.apache.flink.streaming.api.scala.DataStream (not serializable) ---! REMOVED METHOD: PUBLIC(-) boolean iterate$default$3() +++ NEW METHOD: PUBLIC(+) org.apache.flink.streaming.api.scala.DataStream setMaxParallelism(int) *** MODIFIED ANNOTATION: scala.reflect.ScalaSignature ``` i.e. it's complaining about `iterate$default$3()`. The problem seems to be that Scala will generate some obfuscated method name for the `iterate()` method and the `@PublicEvolving` annotation is not properly applied to that. I tried playing around with the japicmp config in the root pom file but to no avail. Maybe you can find something that works. What I added is this: ``` org.apache.flink.streaming.api.scala.DataStream#iterate$default$3() ``` but it seems that doesn't work. ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user mlipkovich commented on the issue: https://github.com/apache/flink/pull/4655 The local build works fine. If you mean `mvn verify` by compatibility plugin it also worked with no issues ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user aljoscha commented on the issue: https://github.com/apache/flink/pull/4655 You can try and run the build locally to have a look at the file generated by the compatibility plugin. Then we can figure out why it's complaining. ---
[GitHub] flink issue #4655: [FLINK-7567]: Removed keepPartitioning parameter from ite...
Github user mlipkovich commented on the issue: https://github.com/apache/flink/pull/4655 As I understand the build has failed because of changed API method. The method which was changed has annotation PublicEvolving so there should be a way to change it. As was mentioned by @aljoscha https://issues.apache.org/jira/browse/FLINK-7567 there is no way to create a method with updated API and to deprecate the current one because of the default parameter ---