parthchandra opened a new issue, #6007:
URL: https://github.com/apache/datafusion-comet/issues/6007
### What is the problem the feature request solves?
With `spark.comet.parquet.write.enabled=true`, a straightforward Parquet
write still runs on Spark when the output path is an object store.
`CometDataWritingCommand.getSupportLevel` rejects anything that isn't `file:`
or `hdfs:`:
```scala
if (!outputPath.startsWith("file:") && !outputPath.startsWith("hdfs:"))
return Unsupported("Supported output filesystems: local, HDFS")
```
Most production Spark writes target object stores (`s3a://`, `gs://`,
`abfss://`), so native write is effectively unavailable there today.
#### Steps to reproduce
```sql
-- comet enabled, spark.comet.parquet.write.enabled=true, incompatible ops
allowed
INSERT OVERWRITE DIRECTORY 's3a://example-bucket/warehouse/out'
USING parquet OPTIONS (compression 'zstd')
SELECT id, name FROM src; -- src is a native Comet scan
```
Explain shows `Execute InsertIntoHadoopFsRelationCommand` staying on Spark;
extended Comet info reports the write operator as unsupported/disabled.
### Describe the potential solution
_No response_
### Additional context
_No response_
--
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]
