Zac Blanco

Add KLL Sketch Functions

5

## Description This change adds KLL Sketch support from the Apache DataSketches library. One of the benefits of KLL sketches is that the implementation supports more datatypes than the existing...

Add histograms for optimizer cost calculation

3

## Description This commits provides two critical changes: 1. Adds a new enum value to ColumnStatisticType: Histogram. 2. Utilizes the new histograms in optimizer's cost calculations ## Histogram SPI Additions...

Fix optimizer performance regression for large IN

4

## Description Fixes the bug found in #22661 This bug occurs due to large IN lists. When computing statistics for queries with this clause, we generate an instance of VariableStatisticEstimate...

Parallelize maven builds by default

2

## Description Adding these parameters defaults all maven builds from this directory to using a parallelized mode if running on a multi-core machine. It can significantly improve build performance when...

Move UI to separate module

5

## Description Moves the UI to its own module which can be built in parallel or excluded entirely from the build in `presto-main`. Fixes #22741 ## Motivation and Context Discussed...

Remove fields from PlanFragment for task serialization

3

## Description This PR removes the `statsAndCosts` field from when a plan fragment is serialized and sent to a worker in a `TaskUpdateRequest`. ## Motivation and Context The PlanFragment class...

## Description This change adds two features: 1. Support for `format ` in EXPLAIN ANALYZE 2. Adds support for a `stats` field in each plan node representation This allows the...

Proactively upload file parts in PrestoS3FileSystem

9

## Description This changes forgoes using the original method of using the AWS SDK `TransferManager` and instead uses the low-level AWS client API in order to support file uploads to...

[Iceberg] Return predicate columns in table statistics

2

## Description Returns `TableStatistics` instances which include `VariableStatsEstimate`s corresponding to the selected columns in the table scan and the columns which are in the table layout predicates. ## Motivation and...

Bytes buffered in memory should be considered written for statistical bookkeeping. This fixes issues where writer scaling doesn't work when there are files smaller than the configured parquet row group...