chucheng92 commented on code in PR #3579: URL: https://github.com/apache/calcite/pull/3579#discussion_r1424776526
########## core/src/main/java/org/apache/calcite/sql/fun/SqlLibraryOperators.java: ########## @@ -1061,23 +1061,29 @@ static RelDataType deriveTypeSplit(SqlOperatorBinding operatorBinding, private static RelDataType arrayReturnType(SqlOperatorBinding opBinding) { final List<RelDataType> operandTypes = opBinding.collectOperandTypes(); - // only numeric & character types check + // only numeric & character types check, this is a special spark array case + // the form like ARRAY(1, 2, '3') will return ["1", "2", "3"] boolean hasNumeric = false; boolean hasCharacter = false; boolean hasOthers = false; for (RelDataType type : operandTypes) { SqlTypeFamily family = type.getSqlTypeName().getFamily(); - requireNonNull(family, "array element type family"); + // some types such as Row, the family is null, fallback to normal inferred type logic + if (family == null) { + hasOthers = true; + break; + } + // skip it because we allow NULL literal + if (SqlTypeUtil.isNull(type)) { + continue; Review Comment: thanks for reviewing. the old code use `case NULL`, the NULL is SqlTypeFamily. however some other types such as UNKNOWN, the family is also NULL. here just use SqlTypeUtil#isNull to truly indicate NULL type. use continue will skip this null literal element just like old logic in switch. -- 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: commits-unsubscr...@calcite.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org