Karol Sobczak

Results 33 issues of Karol Sobczak

When all SortedRangeSets are discrete, then performance of `SortedRangeSet#union` could de improved greatly. Improving union would reduce DF collection latency and reduce coordinator CPU usage (DF are unioned on coordinator)....

good first issue
performance

## Description Also, fix mayFail implementation for Cast ## Additional context and related issues ## Release notes ( ) This is not user-visible or is docs only, and no release...

bug
cla-signed

Currently, we use inequality joins only for expressions like `probe_symbol < build_symbol AND probe_symbol + 1 > build_symbol`, but we don't inequality join for `probe_symbol build_symbol`. Affected queries: ``` tpch/q21...

performance

https://github.com/trinodb/trino/pull/21888 introduces subquery cache feature to Trino engine. However, there are many follow up items to improve the performance and hit rate. Here are the roadmap items: ```[tasklist] ### Tasks...

subquery-cache

Support for joins in subquery cache is important in order to cover the most frequent operators in source stages. The typical use case where join is in source stage is...

subquery-cache

As part of the task `limit` property might need to be added to `PlanSignature`. Alternatively `CacheManager` could check if stored pages have sufficient number of rows

subquery-cache

subqueries like 1. `x = 10 AND y = 10` 2. `y = 10 and x = 10` should have same canonical representation. Order canonicalized conjuncts in some deterministic way.

subquery-cache

Add `ttl` to `CacheSplitId`. This will allow to easily extend cache to JDBC and other ttl based connectors. It will also improve cache utilization as not every column for same...

subquery-cache

General idea is too prioritize cached split data + add tts, so that more frequently used data (based on query history or some other signal) is cached while less frequently...

subquery-cache

Subquery `s1` and `s2` might have common subquery extracted. However, `s1` and `s2` might have limited number of common splits. Therefore it doesn’t make sense for stage `s1` to wait...

subquery-cache