sunchao commented on a change in pull request #2276:
URL: https://github.com/apache/iceberg/pull/2276#discussion_r605091501
##########
File path: spark/src/main/java/org/apache/iceberg/spark/SparkReadOptions.java
##########
@@ -33,6 +33,9 @@ private SparkReadOptions() {
// A timestamp in milliseconds; the snapshot used will be the snapshot
current at this time.
public static final String AS_OF_TIMESTAMP = "as-of-timestamp";
+ // Whether to combine splits by partition boundary or not.
+ public static final String BY_PARTITION = "by-partition";
Review comment:
This is discussed before (see
[here](https://github.com/apache/iceberg/pull/2276#discussion_r588798684)), and
I'm fine for having a table property which is only picked up by Spark at least
initially. Otherwise, we can choose to always do task combining for scan from
Spark (when the table is partitioned).
A better approach, though, is to pass down the info of whether the task
combining is actually utilized in the query. This may require some API change
from Spark.
--
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
