Closed nc6 closed 3 years ago
That's not a regression in sparkle, but a change in upstream Spark. It could be argued that the upstream change is a regression, but I'm not sure. In any case, the point is that Spark 2.2 now requires explicitly specifying extra JAR's to spark-submit
in order to work with AWS. CI does just that. See https://github.com/tweag/sparkle/blob/master/.circleci/config.yml#L61 and the comment just above it.
This example has been running successfully in CI after recent fixes. Closing.
Moving from commit 3c4eccf to f22331f causes the following regression when attempting to run
stack exec -- spark-submit --master 'local[1]' sparkle-example-lda.jar
.Last known good: 3c4eccf First known bad: d0dcbb1
Tested using
stack build --nix
.