nastra commented on code in PR #16363:
URL: https://github.com/apache/iceberg/pull/16363#discussion_r3529805024
##########
core/src/main/java/org/apache/iceberg/TableProperties.java:
##########
@@ -176,14 +176,21 @@ private TableProperties() {}
"write.delete.parquet.row-group-check-max-record-count";
public static final int PARQUET_ROW_GROUP_CHECK_MAX_RECORD_COUNT_DEFAULT =
10000;
- public static final String PARQUET_ROW_GROUP_SIZE_TRACK_UNCOMPRESSED =
- "write.parquet.row-group-size-track-uncompressed";
- public static final boolean
PARQUET_ROW_GROUP_SIZE_TRACK_UNCOMPRESSED_DEFAULT = false;
-
public static final String PARQUET_BLOOM_FILTER_MAX_BYTES =
"write.parquet.bloom-filter-max-bytes";
public static final int PARQUET_BLOOM_FILTER_MAX_BYTES_DEFAULT = 1024 * 1024;
+ // Enables parquet-mr's adaptive bloom filter sizing (PARQUET-2254). When
true,
+ // the writer evaluates ~5 candidate bloom filters and picks the smallest
that
+ // satisfies actual NDV at FPP. Works in combination with
+ // PARQUET_BLOOM_FILTER_MAX_BYTES, which bounds the largest candidate — the
+ // adaptive size never exceeds it. Significantly reduces file size for
+ // low-row-count writes (e.g., streaming microbatches) that would otherwise
pad
+ // the parquet file with an empty `bloom-filter-max-bytes` buffer.
Review Comment:
```suggestion
// the parquet file with an empty PARQUET_BLOOM_FILTER_MAX_BYTES buffer.
```
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
