****************
July 2023 Update
****************

Documentation
=============

* Add :doc:`I/O Optimizations and the TpchBenchmark </develop/TpchBenchmark>` article.
* Add `array_sort lambda function <https://velox-lib.io/blog/array-sort>`_ blog post.
* Add documentation for the new :doc:`CompactRow </develop/serde>` row-wise serialization format.
* Expand documentation for :doc:`aggregate functions with variable-width accumulators </develop/aggregate-functions>`.

Core Library
============

* Add support for aggregations over distinct inputs: `SELECT sum(distinct a)`.
* Add TableWriteMerge operator. :issue:`5546`
* Add :doc:`CompactRow </develop/serde>` row-wise serialization format to optimize shuffle for Presto-on-Spark.
* Convert DATE to logical type backed by INTEGER physical type.
* Enhance Aggregate and Window functions APIs to provide access to configuration properties.
* Add support for compressing spill data. :pr:`5544`

Presto Functions
================

* Add :func:`array_sort_desc`, `array_union` functions.
* Add :func:`array_sort` lambda function.
* Add :func:`cauchy_cdf`, :func:`chi_squared_cdf`, :func:`f_cdf`, :func:`inverse_beta_cdf` function.
* Add :func:`levenshtein_distance` function.
* Add 2-arg versions of :func:`min` and :func:`max` aggregate functions.
* Add :func:`multimap_agg` aggregate function.
* Add :func:`entropy` aggregate function.
* Add support for BOOLEAN comparison argument to :func:`min_by` and :func:`max_by`.
* Add support for IGNORE NULLs to Window value functions.
* Optimize :func:`map_agg` function for inputs with many duplicate keys.
* Optimize :func:`array_agg`, :func:`map_agg`, :func:`set_agg`, :func:`min`, :func:`max` for abandon-partial-aggregation use case.
* Optimize :func:`json_extract`, :func:`json_extract_scalar`, :func:`json_size` using SIMD.

Spark Functions
===============

* Add :spark:func:`acos`, :spark:func:`cosh`, :spark:func:`hypot`, :spark:func:`left`, :spark:func:`log1p`,
  :spark:func:`lpad`, :spark:func:`rpad`, :spark:func:`sinh`, :spark:func:`translate` functions.
* Fix return type of :spark:func:`size` function.
* Fix intermediate type of :spark:func:`first` and :spark:func:`last` functions.

Hive Connector
==============

* Add support for `Tencent Cloud Object Storage <https://www.tencentcloud.com/products/cos>`_. :pr:`5619`
* Add support for reading from `Google Cloud Storage <https://cloud.google.com/storage>`_. :pr:`5069`

Performance and Correctness
===========================

* Optimize round-robin exchange. :pr:`5651`
* Optimize try_cast expression for the case when many rows fail the cast.

Credits
=======

Aditi Pandit, Amit Dutta, Arpit Porwal, Austin Dickey, Bikramjeet Vig,
Chad Austin, Chen Zhang, Chengcheng Jin, Christian Clauss, Christian Zentgraf,
Deepak Majeti, Eric Liu, GOPU-Krishna-S, Ge Gao, Gufan Yin, Harsha Rastogi,
Iosif Spulber, Jacob Wujciak-Jens, Jialiang Tan, Jiayan Wei, Jimmy Lu,
Jin Chengcheng, John Elliott, Kapil Sharma, Karteekmurthys, Ke, Kevin Wilfong,
Konstantin Tsoy, Krishna Pai, Laith Sakka, Logan Wendholt, Ma-Jian1,
Masha Basmanova, Orri Erling, PHILO-HE, Paul Amonson, Pedro Pedreira, Pramod,
Pranjal Shankhdhar, Sandino Flores, Srikrishna Gopu, Srinivas S T, Taras Galkovskyi,
Tiziano Carotti, Wei He, Xiaoxuan Meng, Yang Zhang, Ying, Zac, Zhaolong Zhu,
Zhongjun Jin, duanmeng, joey.ljy, kayagokalp, mayan, mohsaka, rrando901,
rui-mo, wypb, xiaodou, xiaoxmeng, yangchuan, yanmin, zhejiangxiaomai, 陈旭