[WIP][GLUTEN-10215][VL] Delta 4 native write support for Spark 4.0#11924
[WIP][GLUTEN-10215][VL] Delta 4 native write support for Spark 4.0#11924malinjawi wants to merge 2 commits intoapache:mainfrom
Conversation
zhztheplayer
left a comment
There was a problem hiding this comment.
@malinjawi looks good.
I think we'd first open and merge a PR which adds tests, then switch to this one?
|
|
||
| import org.apache.spark.sql.execution.datasources.v2.OffloadDeltaCommand | ||
|
|
||
| class VeloxDelta40WriteComponent extends Component { |
There was a problem hiding this comment.
A out-of-topic question: Do you think we can possibly support e.g., src-spark4 / src-delta4 folder-activation in Gluten's Maven settings? So we might not have to port these code again for Spark 4.1 if lucky.
Something like:
backends-velox
|- src-spark3
|- src-spark34
|- src-spark35
|- src-spark4
|- src-spark40
|- src-spark41
|- src-delta
|- src-delta4
|- src-delta40
|- src-delta41
...
Current code is a bit inflexible, it was introduced in #9996.
|
|
||
| import org.apache.spark.sql.execution.datasources.v2.OffloadDeltaCommand | ||
|
|
||
| class VeloxDelta40WriteComponent extends Component { |
There was a problem hiding this comment.
A component file is needed to activate the component: https://github.com/apache/gluten/blob/main/backends-velox/src-delta33/main/resources/META-INF/gluten-components/org.apache.gluten.component.VeloxDelta33WriteComponent
Hey @zhztheplayer! Agreed I had added the UT for this path in separate branches. I am raising the base delta 4.0 UT here: #11938 Also planned to expand testing as needed in separate patches. |
What changes are proposed in this pull request?
This PR adds native Delta Lake write support for Spark 4.0 + Delta 4.0 in the Velox backend.
Main changes:
Scope note:
Part of #10215.
How was this patch tested?