[
https://issues.apache.org/jira/browse/SPARK-57807?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Max Gekk updated SPARK-57807:
-----------------------------
Description:
This sub-task is part of the umbrella SPARK-56822 (timestamps with nanosecond
precision). It documents a known limitation rather than adding support.
h2. Decision
Document only, not implement (per review).
h2. Problem
SparkR maps only {{"timestamp"}} to {{POSIXct}} in {{PRIMITIVE_TYPES}}
(R/pkg/R/types.R ~L21-33); {{schema.R}} / {{DataFrame.R}} collect through
{{PRIMITIVE_TYPES}}, with no awareness of {{TimestampNTZNanosType}} /
{{TimestampLTZNanosType}}. As a result, nanosecond-precision timestamp columns
cannot round-trip through SparkR schema/collect. Per project decision, SparkR
support is *not* being added.
h2. Goal
Document the limitation: nanosecond-precision timestamp types
({{TIMESTAMP_NTZ(p)}} / {{TIMESTAMP_LTZ(p)}}, {{p}} in [7, 9]) are *not
supported* in SparkR (schema and collect), and describe the behavior when such
a column is encountered.
h2. Scope
Docs only: add a note to the SparkR documentation / known-limitations section,
cross-linked with the nanosecond-timestamp migration guide and the typed-API
limitation note. No changes to the R type mapping.
h2. Acceptance criteria
* SparkR docs clearly state that nanosecond-precision timestamp types are
unsupported (schema and collect).
h2. Testing
Docs only.
> Document that SparkR does not support nanosecond-precision timestamp types
> --------------------------------------------------------------------------
>
> Key: SPARK-57807
> URL: https://issues.apache.org/jira/browse/SPARK-57807
> Project: Spark
> Issue Type: Sub-task
> Components: Documentation
> Affects Versions: 5.0.0
> Reporter: Max Gekk
> Priority: Major
>
> This sub-task is part of the umbrella SPARK-56822 (timestamps with nanosecond
> precision). It documents a known limitation rather than adding support.
> h2. Decision
> Document only, not implement (per review).
> h2. Problem
> SparkR maps only {{"timestamp"}} to {{POSIXct}} in {{PRIMITIVE_TYPES}}
> (R/pkg/R/types.R ~L21-33); {{schema.R}} / {{DataFrame.R}} collect through
> {{PRIMITIVE_TYPES}}, with no awareness of {{TimestampNTZNanosType}} /
> {{TimestampLTZNanosType}}. As a result, nanosecond-precision timestamp
> columns cannot round-trip through SparkR schema/collect. Per project
> decision, SparkR support is *not* being added.
> h2. Goal
> Document the limitation: nanosecond-precision timestamp types
> ({{TIMESTAMP_NTZ(p)}} / {{TIMESTAMP_LTZ(p)}}, {{p}} in [7, 9]) are *not
> supported* in SparkR (schema and collect), and describe the behavior when
> such a column is encountered.
> h2. Scope
> Docs only: add a note to the SparkR documentation / known-limitations
> section, cross-linked with the nanosecond-timestamp migration guide and the
> typed-API limitation note. No changes to the R type mapping.
> h2. Acceptance criteria
> * SparkR docs clearly state that nanosecond-precision timestamp types are
> unsupported (schema and collect).
> h2. Testing
> Docs only.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
