[SPARK-20361] JVM locale affects SQL type names

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 2.1.0
• Component/s: PySpark, SQL
• External issue URL: http://stackoverflow.com/q/43455592
• Assignee: Unassigned
• Reporter: Maciej Szymkiewicz
• Created: 17/Apr/17 21:49
• Updated: 12/Dec/22 18:11
• Resolved: 18/Apr/17 09:59

Description

Steps to reproduce:

from pyspark.sql.types import IntegerType

locale = sc._jvm.java.util.Locale
locale.setDefault(locale.forLanguageTag("tr-TR"))

spark.createDataFrame([1, 2, 3], IntegerType())

Py4JJavaError: An error occurred while calling o24.applySchemaToPythonRDD.
: java.util.NoSuchElementException: key not found: integer
	at scala.collection.MapLike$class.default(MapLike.scala:228)
...

Issue Links

• duplicates: SPARK-20156 Java String toLowerCase "Turkish locale bug" causes Spark problems