echauchot commented on a change in pull request #15529:
URL: https://github.com/apache/beam/pull/15529#discussion_r719126109
##########
File path: runners/spark/spark_runner.gradle
##########
@@ -166,6 +166,10 @@ dependencies {
provided
"org.apache.spark:spark-network-common_$spark_scala_version:$spark_version"
provided "org.apache.spark:spark-sql_$spark_scala_version:$spark_version"
provided
"org.apache.spark:spark-streaming_$spark_scala_version:$spark_version"
+ runtimeOnly "org.apache.spark:spark-core_$spark_scala_version:$spark_version"
Review comment:
> 1. Why do we need to have two configurations (`provided` and
`runtimeOnly`) for the same deps?
> 2. Can you make sure that generated jars won't cause breaking changes for
users who already use any of Spark Runners?
1. because provided scope is given by a gradle plugin and is not equivalent
to maven provided scope, it lacks putting the libs in the runtime classpath.
They are used for local in memory runs for example Nexmark. In gradle we can
extend the scopes by adding multiple scope lines.
2. I Like I said in the description of the PR I checked that the generated
jars do not contain apache spark libs.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
