[
https://issues.apache.org/jira/browse/SPARK-59974?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nikolina Vraneš updated SPARK-59974:
------------------------------------
Affects Version/s: 4.4.0
(was: 5.0.0)
> Reduce distance evaluations in nearest ASOF joins
> -------------------------------------------------
>
> Key: SPARK-59974
> URL: https://issues.apache.org/jira/browse/SPARK-59974
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 4.4.0
> Reporter: Nikolina Vraneš
> Priority: Major
>
> Nearest ASOF joins evaluate the distance expression for each eligible row in
> a sorted right-side group. For integral and timestamp keys, only the closest
> eligible rows on either side of the left key need comparison.
> Compare these boundary candidates with exact arithmetic to avoid integer
> overflow. Compare timestamps by elapsed UTC time, including full nanosecond
> precision. Keep the distance-based scan for fractional distances, where
> rounding can affect ties.
> The tests cover duplicate keys, overflow with ANSI mode on and off, a
> daylight saving transition, nanosecond precision, residual predicates,
> spills, and floating-point ties.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
