There is a performance regression in
ROUND()to prevent returning incorrect results due to integer / double overflows.
Fix the compilation error when aggregation has order by clause and the input is a function.
IF(predicate, AGG(x))to aggregation with mask in plan level. This is controlled by system property
optimize_conditional_aggregation_enabledand defaults to false.
Add new security API
selectAuthorizedIdentityand new configuration property
selectAuthorizedIdentityprevents potential security loopholes, e.g., reading illegal data from a fake username.
Add memory limit check for HashBuilderOperator during memory revoke.
Add null masking for the Parquet decryption feature. When this feature is enabled and the user is denied access encrypted column, the columns will be removed in the requested schema sent to Parquet. Then it is filled out with
NULLwhen the result is returned.
Add optimization for
approx_percentile()functions evaluation. Multiple
approx_percentile()functions on the same field will be combined into one
approx_percentile()function which takes an array of percentile as arguments. The optimization is controlled by session property
optimize_multiple_approx_percentile_on_same_fieldwhich is true by default.
Add optimization for outer join by add randomized value for NULL join keys to avoid skew in NULL. This optimization is turned off by default and can be turned on by setting
Add retry with increased partition count if query fails due to low partition count. This can be enabled with the
map_subset(). This function takes a map and an array of keys and returns a map with entries from the input map with keys contained in the array supplied.
Upgrade Apache Iceberg version from 0.14.0 to 0.14.1.
Upgrade Java Topology Suite (jts) library version to 1.19.0.
Delta Lake Connector Changes#
Improve performance of reading newly created tables.
CREATE TABLEsupport to Delta connector.
DROP TABLEsupport for the external table to Delta connector.
Filesystem Connector Changes#
Add support for the HDFS filesystem connector in Presto Native Execution.
Tpc-h Connector Changes#
Add support for the TPC-H connector in Presto Native Execution. Velox only supports standard column naming. The tpch connector property
tpch.column-naming=standardmust be set on the Java side.
ErrorTypefrom presto-spi to presto-common.
Fix the issue which causes query failures when the Parquet file statistics is corrupted.
Add a new session property
read_null_masked_parquet_encrypted_value_enabledto toggle the Parquet null masking feature. This session property defaults to false.
Upgrade the Apache Hudi version to 0.12.0.
BYTEStype decoding issue.
Add new config
pinot.query-optionsand session property
pinot.query_optionsto set [Pinot Query Options](https://docs.pinot.apache.org/users/user-guide-query/query-options) for generated Pinot query.
Remove catalog config:
pinot.use-pinot-sql-for-broker-queriesand session config:
Remove unused configs:
Remove unused session configs:
Deprecate Pinot PQL query endpoint, by default using SQL query endpoint.
Deprecate Pinot netty server query.
Upgrade Pinot release version to 0.11.0.
Add the weighted round-robin scheduling in the router.
Aditi Pandit, Ahmed ElSherbiny, Ajay George, Amit Dutta, Amr Elroumy, Arjun Gupta, Arunachalam Thirupathi, Behnam Robatmili, Beinan, Chen Yang, Chunxu Tang, Deepak Majeti, Feilong Liu, Ge Gao, James Sun, Jimmy Lu, Karteek Murthy Samba Murthy, Krishna Pai, Lin Liu, MJ Deng, Masha Basmanova, Michael Shang, Milosz Linkiewicz, Naresh Kumar, Naveen Kumar Mahadevuni, Neerad Somanchi, Nizar Hejazi, Pranjal Shankhdhar, Rebecca Schlussel, Reetika Agrawal, Robert Stupp, Rohit Jain, Sacha Viscaino, Sagar Sumit, Sergey Pershin, Sergii Druzkin, Sreeni Viswanadha, Swapnil Tailor, Timothy Meehan, Todd Gao, Xiang Fu, Xinli Shang, Y Ethan Guo, abhiseksaikia, dnskr, pratyakshsharma, singcha, tanjialiang, xiaoxmeng, yingsu00