12.16. Release 318 (26 Aug 2019)
General Changes
- Fix query failure when using
DISTINCT FROM with the UUID or
IPADDRESS types. (#1180)
- Improve query performance when
optimize_hash_generation is enabled. (#1071)
- Improve performance of information schema tables. (#999, #1306)
- Rename
http.server.authentication.* configuration options to http-server.authentication.*. (#1270)
- Change query CPU tracking for resource groups to update periodically while
the query is running. Previously, CPU usage would only update at query
completion. This improves resource management fairness when using
CPU-limited resource groups. (#1128)
- Remove
distributed_planning_time_ms column from system.runtime.queries. (#1084)
- Add support for
Asia/Qostanay time zone. (#1221)
- Add session properties that allow overriding the query per-node memory limits:
query_max_memory_per_node and query_max_total_memory_per_node. These properties
can be used to decrease limits for a query, but not to increase them. (#1212)
- Add Google Sheets Connector. (#1030)
- Add
planning_time_ms column to the system.runtime.queries table that shows
the time spent on query planning. This is the same value that used to be in the
analysis_time_ms column, which was a misnomer. (#1084)
- Add
last_day_of_month() function. (#1295)
- Add support for cancelling queries via the
system.runtime.kill_query procedure when
they are in the queue or in the semantic analysis stage. (#1079)
- Add queries that are in the queue or in the semantic analysis stage to the
system.runtime.queries table. (#1079)
Web UI Changes
- Display information about queries that are in the queue or in the semantic analysis
stage. (#1079)
- Add support for cancelling queries that are in the queue or in the semantic analysis
stage. (#1079)
Hive Connector Changes
- Fix query failure due to missing credentials while writing empty bucket files. (#1298)
- Fix bucketing of
NaN values of real type. Previously NaN values
could be assigned a wrong bucket. (#1336)
- Fix reading
RCFile collection delimiter set by Hive version earlier than 3.0. (#1321)
- Return proper error when selecting
"$bucket" column from a table using
Hive bucketing v2. (#1336)
- Improve performance of S3 object listing. (#1232)
- Improve performance when reading data from GCS. (#1200)
- Add support for reading data from S3 Requester Pays buckets. This can be enabled
using the
hive.s3.requester-pays.enabled configuration property. (#1241)
- Allow inserting into bucketed, unpartitioned tables. (#1127)
- Allow inserting into existing partitions of bucketed, partitioned tables. (#1347)
PostgreSQL Connector Changes
- Add support for providing JDBC credential in a separate file. This can be enabled by
setting the
credential-provider.type=FILE and connection-credential-file
config options in the catalog properties file. (#1124)
- Allow logging all calls to
JdbcClient. This can be enabled by turning
on DEBUG logging for io.prestosql.plugin.jdbc.JdbcClient. (#1274)
- Add possibility to force mapping of certain types to
varchar. This can be enabled
by setting jdbc-types-mapped-to-varchar to comma-separated list of type names. (#186)
- Add support for PostgreSQL
timestamp[] type. (#1023, #1262, #1328)
MySQL Connector Changes
- Add support for providing JDBC credential in a separate file. This can be enabled by
setting the
credential-provider.type=FILE and connection-credential-file
config options in the catalog properties file. (#1124)
- Allow logging all calls to
JdbcClient. This can be enabled by turning
on DEBUG logging for io.prestosql.plugin.jdbc.JdbcClient. (#1274)
- Add possibility to force mapping of certain types to
varchar. This can be enabled
by setting jdbc-types-mapped-to-varchar to comma-separated list of type names. (#186)
Redshift Connector Changes
- Add support for providing JDBC credential in a separate file. This can be enabled by
setting the
credential-provider.type=FILE and connection-credential-file
config options in the catalog properties file. (#1124)
- Allow logging all calls to
JdbcClient. This can be enabled by turning
on DEBUG logging for io.prestosql.plugin.jdbc.JdbcClient. (#1274)
- Add possibility to force mapping of certain types to
varchar. This can be enabled
by setting jdbc-types-mapped-to-varchar to comma-separated list of type names. (#186)
SQL Server Connector Changes
- Add support for providing JDBC credential in a separate file. This can be enabled by
setting the
credential-provider.type=FILE and connection-credential-file
config options in the catalog properties file. (#1124)
- Allow logging all calls to
JdbcClient. This can be enabled by turning
on DEBUG logging for io.prestosql.plugin.jdbc.JdbcClient. (#1274)
- Add possibility to force mapping of certain types to
varchar. This can be enabled
by setting jdbc-types-mapped-to-varchar to comma-separated list of type names. (#186)
SPI Changes
- Add
Block.isLoaded() method. (#1216)
- Update security APIs to accept the new
ConnectorSecurityContext
and SystemSecurityContext classes. (#171)
- Allow connectors to override minimal schedule split batch size. (#1251)