flyrain commented on code in PR #4588:
URL: https://github.com/apache/iceberg/pull/4588#discussion_r950762704
##########
spark/v3.3/spark/src/main/java/org/apache/iceberg/spark/data/vectorized/ColumnarBatchReader.java:
##########
@@ -183,6 +187,8 @@ Pair<int[], Integer>
buildPosDelRowIdMapping(PositionDeleteIndex deletedRowPosit
currentRowId++;
} else if (hasIsDeletedColumn) {
isDeleted[originalRowId] = true;
+ } else {
+ deletes.incrementDeleteCount();
}
Review Comment:
It is triggered by either projection or filtering. Here are two examples:
```
select name, _deleted from student
```
```
select name from student where _deleted=true
```
It is true that user can get row-level deleted rows, users can also get both
deleted and un-deleted rows with `_deleted` column, the first sql above returns
that. The way I saw it is that the delete count metric is orthogonal to the
metadata column `_deleted`. It reports how many row-level deletes applied as
its label indicates, `public static final String DISPLAY_STRING = "number of
row deletes applied"`. I don’t have strong option on it though. What do you
think? BTW, the non-vectorized read also didn’t report the metric in case of
`_deleted` column, see line 255 in class `DeleteFilter`.
--
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]
