12.14. Release 321 (15 Oct 2019)#
Warning
The server RPM is broken in this release.
General Changes#
- Fix incorrect result of
round()
when applied to atinyint
,smallint
,integer
, orbigint
type with negative decimal places. (#42) - Improve performance of queries with
LIMIT
overinformation_schema
tables. (#1543) - Improve performance for broadcast joins by using dynamic filtering. This can be enabled
via the
experimental.enable-dynamic-filtering
configuration option or theenable_dynamic_filtering
session property. (#1686)
Hive Connector Changes#
- Fix reading
TEXT
file collection delimiter set by Hive versions earlier than 3.0. (#1714) - Fix a regression that prevented Presto from using the AWS Glue metastore. (#1698)
- Allow skipping header or footer lines for
CSV
format tables via theskip_header_line_count
andskip_footer_line_count
table properties. (#1090) - Rename table property
textfile_skip_header_line_count
toskip_header_line_count
andtextfile_skip_footer_line_count
toskip_footer_line_count
. (#1090) - Add support for LZOP compressed (
.lzo
) files. Previously, queries accessing LZOP compressed files would fail, unless all files were small. (#1701) - Add support for bucket-aware read of tables using bucketing version 2. (#538)
- Add support for writing to tables using bucketing version 2. (#538)
- Allow caching directory listings for all tables or schemas. (#1668)
- Add support for dynamic filtering for broadcast joins. (#1686)