[SPARK-38011] Remove useless and duplicated configuration in ParquetFileFormat builderReader

• Type: Improvement
• Status: Resolved
• Priority: Trivial
• Resolution: Fixed
• Affects Version/s: 3.2.0
• Fix Version/s: 3.3.0
• Component/s: SQL
• Assignee: yikaifei
• Reporter: yikaifei
• Created: 25/Jan/22 03:11
• Updated: 27/Jan/22 07:56
• Resolved: 27/Jan/22 07:56

Description

Currently, we called `ParquetWriteSupport.setSchema(requiredSchema, hadoopConf)` when ParquetFileFormat buildReaderWithPartition, and this helper method set `org.apache.spark.sql.parquet.row.attributes` and `parquet.writer.version`, actually,  the former is already set and latter is useless when parquetRead

Issue Links

• links to: [Github] Pull Request #35308 (Yikf)