Robert Stupp
Robert Stupp
I tried the `spotlessPredeclare` approach [here](https://github.com/snazy/cel-java/commit/0011fb63f32a164124a32b27213dc3ea085fc28e), but it sadly didn't change the outcome. In fact, the `spotlessPredeclare` adds another variant, so there are now 2 spotless variants.
Ran into the same issue w/ latest spotless plugin version 6.8.0 in another way (https://github.com/projectnessie/nessie/issues/4655). The change that caused the issue to happen was an unrelated dependency version bump.
Hm. Seems this is caused by Spark and can be controlled via the configuration `spark.sql.defaultUrlStreamHandlerFactory.enabled`. See `org.apache.spark.sql.internal.SharedState#setFsUrlStreamHandlerFactory`
Once Nessie can at least use Java 11's new HTTP client with #4448 , the problem will not happen at all when using Java 11 or newer.
@dimas-b FYI
Can you re-try and do the "delete" in the commit right before the commit that contains the key-list? The key-list is generated for all the commits _before_ the commit w/...
Well, my ask was to verify the key-list generated _after_ the commit with the delete-op. Deleted keys don’t make it into the aggregated key-list.
We could revisit this one for v2 API as well.
> just looks like nothing may have been tested (or at least i didnt find a test summary or the like) all actual tests are integration tests - via 'intTest`
Nothing to migrate yet, so removing the 1.0.0 milestone