andygrove opened a new pull request, #6551:
URL: https://github.com/apache/datafusion-comet/pull/6551
## Which issue does this PR close?
No issue. This is a docs-only change.
## Rationale for this change
The docs don't say which Spark plan nodes Comet leaves in place on purpose,
or why. The "Not currently planned" list on the operators page covers operators
that do real work but aren't on the roadmap. The wrapper list in Understanding
Comet Plans only explains the coverage counts. The typed `Dataset` operators
aren't mentioned anywhere.
Without this written down, a contributor can spend time on a serde for a
node that does no work, such as `ResultQueryStageExec`. A user can also read a
`ResultQueryStage` in a plan as a fallback.
## What changes are included in this PR?
- `contributor-guide/adding_a_new_operator.md`: a new section, "Operators
That Should Not Be Converted", covering three groups:
- wrappers and scheduling nodes (AQE query stages including
`ResultQueryStageExec`, `AQEShuffleReadExec`, `InputAdapter`,
`WholeStageCodegenExec`, reuse markers)
- typed `Dataset` operators (`DeserializeToObjectExec`, `MapElementsExec`,
`SerializeFromObjectExec`, and the rest)
- driver-side commands (`ExecutedCommandExec`)
It ends with what to do when a new Spark version adds a wrapper node.
- `user-guide/latest/operators.md`: a "Wrapper nodes" section saying these
nodes are not fallbacks, and a "Typed Dataset operators" entry under "Not
currently planned".
## How are these changes tested?
Docs only. `prettier --check` passes on both files. I checked the class
names and the behavior described against the Spark 4.0.4 source
(`objects.scala`, `SparkStrategies`, `AdaptiveSparkPlanExec`, `QueryStageExec`)
and against `CometExecRule` and `ExtendedExplainInfo`.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
