Github user MaxGekk commented on a diff in the pull request: https://github.com/apache/spark/pull/23132#discussion_r237091555 --- Diff: docs/sql-migration-guide-upgrade.md --- @@ -9,6 +9,8 @@ displayTitle: Spark SQL Upgrading Guide ## Upgrading From Spark SQL 2.4 to 3.0 + - In Spark version 2.4 and earlier, accepted format of decimals parsed from JSON is an optional sign ('+' or '-'), followed by a sequence of zero or more decimal digits, optionally followed by a fraction, optionally followed by an exponent. Any commas were removed from the input before parsing. Since Spark 3.0, format varies and depends on locale which can be set via JSON option `locale`. The default locale is `en-US`. To switch back to previous behavior, set `spark.sql.legacy.decimalParsing.enabled` to `true`. --- End diff -- > Is there any performance regression with DecimalFormat? I haven't benchmarked the changes. Looking at `JSONBenchmarks`, we even don't check decimals there. > Do we need the DecimalFormat when locale is en-US? No, we don't need it.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org