12.2. Release 333 (04 May 2020)#
General Changes#
- Fix planning failure when lambda expressions are repeated in a query. (#3218)
- Fix failure when input to
TRY
is a constantNULL
. (#3408) - Fix failure for SHOW CREATE TABLE for tables with row types that contain special characters. (#3380)
- Fix failure when using
max_by()
ormin_by()
where the second argument is of typevarchar
. (#3424) - Fix rare failure due to an invalid size estimation for T-Digests. (#3625)
- Do not require coordinator to have spill paths setup when spill is enabled. (#3407)
- Improve performance when dynamic filtering is enabled. (#3413)
- Improve performance of queries involving constant scalar subqueries (#3432)
- Allow overriding the count of available workers used for query cost
estimation via the
cost_estimation_worker_count
session property. (#2705) - Add data integrity verification for Presto internal communication. This can be configured
with the
exchange.data-integrity-verification
configuration property. (#3438) - Add support for
LIKE
predicate to SHOW COLUMNS. (#2997) - Add SHOW CREATE SCHEMA. (#3099)
- Add
starts_with()
function. (#3392)
Server Changes#
- Require running on Java 11 or above. (#2799)
BigQuery Connector Changes#
Cassandra Connector Changes#
- Allow INSERT statement for table having hidden
id
column. (#3499) - Add support for CREATE TABLE statement. (#3478)
Hive Connector Changes#
- Fix incorrect query results when reading Parquet data with a
varchar
column predicate which is a comparison with a value containing non-ASCII characters. (#3517) - Ensure cleanup of resources (file descriptors, sockets, temporary files, etc.) when an error occurs while writing an ORC file. (#3390)
- Generate multiple splits for files in bucketed tables. (#3455)
- Make file system caching honor Hadoop properties from
hive.config.resources
. (#3557) - Disallow enabling file system caching together with S3 security mapping or GCS access tokens. (#3571)
- Disable file system caching parallel warmup by default. It is currently broken and should not be enabled. (#3591)
- Include metrics from S3 Select in the S3 JMX metrics. (#3429)
- Report timings for request retries in S3 JMX metrics. Previously, only the first request was reported. (#3429)
- Add S3 JMX metric for client retry pause time (how long the thread was asleep between request retries in the client itself). (#3429)
- Add support for SHOW CREATE SCHEMA. (#3099)
- Add
hive.projection-pushdown-enabled
configuration property andprojection_pushdown_enabled
session property. (#3490) - Add support for connecting to the Thrift metastore using TLS. (#3440)