Max Gekk created SPARK-57821:
--------------------------------
Summary: Support nanosecond-precision timestamps in date_trunc
Key: SPARK-57821
URL: https://issues.apache.org/jira/browse/SPARK-57821
Project: Spark
Issue Type: Sub-task
Components: SQL
Affects Versions: 4.3.0
Reporter: Max Gekk
This sub-task is part of the umbrella SPARK-56822 (timestamps with nanosecond
precision).
h2. Problem
{{TruncTimestamp}} (datetimeExpressions.scala ~L2874-2900) declares
{{inputTypes = Seq(StringType, TimestampType)}} and {{dataType =
TimestampType}}, and evaluates via {{DateTimeUtils.truncTimestamp(Long, ...)}}.
A {{TIMESTAMP_NTZ(p)}} / {{TIMESTAMP_LTZ(p)}} argument is implicitly cast to
microsecond {{TimestampType}}, dropping {{nanosWithinMicro}} and forcing an LTZ
result even for NTZ input. ({{trunc(date, ...)}} is date-only and out of scope.)
h2. Goal
{{date_trunc(fmt, ts_nanos)}} accepts nanosecond timestamps, returns the same
nanosecond type/precision and family, and truncates correctly (units at or
above SECOND zero the whole fraction including {{nanosWithinMicro}};
MICROSECOND zeroes {{nanosWithinMicro}}).
h2. Scope
Extend {{TruncTimestamp.inputTypes}} / {{dataType}} to accept
{{AnyTimestampNanoType}}; add a nanosecond eval/codegen branch operating on
{{TimestampNanosVal}}; add {{DateTimeUtils.truncTimestampNanos}} if needed;
keep the format-unit list unchanged (optionally accept a no-op NANOSECOND unit).
h2. Acceptance criteria
* {{date_trunc('SECOND'|'MINUTE'|'HOUR'|'DAY'|...)}} over NTZ/LTZ nanos returns
the same type with the fraction zeroed; {{date_trunc('MICROSECOND', ...)}}
zeroes only {{nanosWithinMicro}}; NTZ stays NTZ, LTZ stays LTZ.
h2. Testing
{{DateExpressionsSuite}}; nanos golden files.
h2. Dependencies
None - independent.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
