cloud-fan commented on a change in pull request #30156:
URL: https://github.com/apache/spark/pull/30156#discussion_r514894838
##########
File path: docs/sql-migration-guide.md
##########
@@ -51,6 +51,8 @@ license: |
- In Spark 3.1, loading and saving of timestamps from/to parquet files fails
if the timestamps are before 1900-01-01 00:00:00Z, and loaded (saved) as the
INT96 type. In Spark 3.0, the actions don't fail but might lead to shifting of
the input timestamps due to rebasing from/to Julian to/from Proleptic Gregorian
calendar. To restore the behavior before Spark 3.1, you can set
`spark.sql.legacy.parquet.int96RebaseModeInRead` or/and
`spark.sql.legacy.parquet.int96RebaseModeInWrite` to `LEGACY`.
- In Spark 3.1, the `schema_of_json` and `schema_of_csv` functions return
the schema in the SQL format in which field names are quoted. In Spark 3.0, the
function returns a catalog string without field quoting and in lower case.
+
+ - In Spark 3.1, when
`spark.sql.legacy.transformationPadNullWhenValueLessThenSchema` is true, Spark
will pad NULL value when script transformation's output value size less then
schema size in default-serde mode(script transformation with row format of `ROW
FORMAT DELIMITED`). If false, Spark will keep original behavior to throw
`ArrayIndexOutOfBoundsException`.
Review comment:
Please follow other migration guide items: first explain what's the
behavior change, then mention how to restore rge legacy behavior with the
legacy config.
##########
File path: docs/sql-migration-guide.md
##########
@@ -51,6 +51,8 @@ license: |
- In Spark 3.1, loading and saving of timestamps from/to parquet files fails
if the timestamps are before 1900-01-01 00:00:00Z, and loaded (saved) as the
INT96 type. In Spark 3.0, the actions don't fail but might lead to shifting of
the input timestamps due to rebasing from/to Julian to/from Proleptic Gregorian
calendar. To restore the behavior before Spark 3.1, you can set
`spark.sql.legacy.parquet.int96RebaseModeInRead` or/and
`spark.sql.legacy.parquet.int96RebaseModeInWrite` to `LEGACY`.
- In Spark 3.1, the `schema_of_json` and `schema_of_csv` functions return
the schema in the SQL format in which field names are quoted. In Spark 3.0, the
function returns a catalog string without field quoting and in lower case.
+
+ - In Spark 3.1, when
`spark.sql.legacy.transformationPadNullWhenValueLessThenSchema` is true, Spark
will pad NULL value when script transformation's output value size less then
schema size in default-serde mode(script transformation with row format of `ROW
FORMAT DELIMITED`). If false, Spark will keep original behavior to throw
`ArrayIndexOutOfBoundsException`.
Review comment:
Please follow other migration guide items: first explain what's the
behavior change, then mention how to restore the legacy behavior with the
legacy config.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
