Github user henryr commented on a diff in the pull request: https://github.com/apache/spark/pull/19769#discussion_r155690933 --- Diff: sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/VectorizedColumnReader.java --- @@ -294,11 +304,22 @@ private void decodeDictionaryIds( break; case INT96: if (column.dataType() == DataTypes.TimestampType) { - for (int i = rowId; i < rowId + num; ++i) { - // TODO: Convert dictionary of Binaries to dictionary of Longs - if (!column.isNullAt(i)) { - Binary v = dictionary.decodeToBinary(dictionaryIds.getDictId(i)); - column.putLong(i, ParquetRowConverter.binaryToSQLTimestamp(v)); + if (!shouldConvertTimestamps()) { + for (int i = rowId; i < rowId + num; ++i) { + if (!column.isNullAt(i)) { + Binary v = dictionary.decodeToBinary(dictionaryIds.getDictId(i)); + column.putLong(i, ParquetRowConverter.binaryToSQLTimestamp(v)); + } + } + } else { + for (int i = rowId; i < rowId + num; ++i) { + // TODO: Convert dictionary of Binaries to dictionary of Longs --- End diff -- Yep, I've removed it because I think the idea needs a bit more discussion re: tradeoffs. I filed [SPARK-22736](https://issues.apache.org/jira/browse/SPARK-22736).
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org