vlsi commented on a change in pull request #2329:
URL: https://github.com/apache/calcite/pull/2329#discussion_r561950919
##########
File path:
linq4j/src/main/java/org/apache/calcite/linq4j/EnumerableDefaults.java
##########
@@ -4683,4 +4683,191 @@ private void flush() {
}
};
}
+
+ /**
+ * Merge Union Enumerable.
+ * Performs a union (or union all) of all its inputs (which must be already
sorted),
+ * respecting the order.
+ *
+ * @param sources input enumerables (must be already sorted)
+ * @param sortKeySelector sort key selector
+ * @param sortComparator sort comparator to decide the next item
+ * @param all whether duplicates will be considered or not
+ * @param equalityComparer {@link EqualityComparer} to control duplicates,
+ * only used if {@code all} is {@code false}
+ * @param <TSource> record type
+ * @param <TKey> sort key
+ */
+ public static <TSource, @Nullable TKey> Enumerable<TSource> mergeUnion(
Review comment:
Currently you declare `sortKeySelector` as `Function1<TSource, TKey>
sortKeySelector`.
In other words, it **must** work for all `TSource` values no matter if
`TSource` is null or not.
On the other hand, if you implement custom `null` behavior, you basically
augment user-provided `sortKeySelector`, and you basically forbid users to
handle `nulls` differently.
I believe this silent implementation detail is worse than a clear NPE in
case the user-provided function fails to handle nulls.
----------------------------------------------------------------
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]
