gatorsmile commented on a change in pull request #23391: [SPARK-26456][SQL]
Cast date/timestamp to string by Date/TimestampFormatter
URL: https://github.com/apache/spark/pull/23391#discussion_r244671504
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/PartitioningUtils.scala
##########
@@ -230,7 +235,7 @@ object PartitioningUtils {
// Once we get the string, we try to parse it and find the partition
column and value.
val maybeColumn =
parsePartitionColumn(currentPath.getName, typeInference,
userSpecifiedDataTypes,
- validatePartitionColumns, timeZone)
+ validatePartitionColumns, timeZone, dateFormatter,
timestampFormatter)
Review comment:
When the partition/bucket column is Date type, our parquet writer convert
the date value to a string and record it as the directory name. Is that
possible that our Spark could return a wrong result?
For example,
- Join two partitioned tables using the partitioned key column whose data
type is Date.
- One table is wrote by Spark 2.4 or prior, and another table is wrote by
Spark 3.0 or later.
- The partition columns contain the value before October 1582.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
