Github user tgravescs commented on a diff in the pull request: https://github.com/apache/spark/pull/22134#discussion_r210910385 --- Diff: sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSource.scala --- @@ -609,7 +609,13 @@ object DataSource extends Logging { /** Given a provider name, look up the data source class definition. */ def lookupDataSource(provider: String, conf: SQLConf): Class[_] = { - val provider1 = backwardCompatibilityMap.getOrElse(provider, provider) match { + val customBackwardCompatibilityMap = + conf.getAllConfs + .filter(_._1.startsWith("spark.sql.datasource.map")) + .map{ case (k, v) => (k.replaceFirst("^spark.sql.datasource.map.", ""), v) } + val compatibilityMap = backwardCompatibilityMap ++ customBackwardCompatibilityMap --- End diff -- so at this point we are leaving com.databricks.spark.avro -> internal avro as the default and users have to set it back to com.databricks.spark.avro or do they set it empty? Although if set to empty I think it will return empty below which will cause an issue. We should have a test case for empty and perhaps have a check for it below in that case. What about documentation? Is there a jira for documenting all avro stuff? If we do leave it as default we definitely want a release note with change in behavior.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org