github-actions[bot] commented on code in PR #64822:
URL: https://github.com/apache/doris/pull/64822#discussion_r3472090211
##########
regression-test/suites/datatype_p2/complex_types/two_level_nestedtypes_with_s3data.groovy:
##########
@@ -112,7 +112,8 @@ suite("two_level_nestedtypes_with_s3data", "p2") {
"provider" = "${getS3Provider()}",
"read_json_by_line"="true") order by c1,c2; """
} else {
- order_qt_sql_s3 """select c_bool, c_double, c_decimal, c_date,
c_char from s3(
+ // Keep direct s3() smoke query scalar; nested complex values are
validated after loading into Doris.
+ order_qt_sql_s3 """select k1 from s3(
Review Comment:
The new smoke query only projects `k1`, so this no longer covers direct
Parquet `s3()` projection of any nested column. The later `insert into
${table_name} select * from s3(...)` plus table queries proves the data can be
loaded and read from Doris storage, but it does not exercise the same direct
TVF query shape that used to return complex slots while ordering by `k1`. That
shape is important because Parquet/ORC `s3` TVFs are eligible for TopN lazy
materialization, so a direct `select <complex expr> from s3(...) order by k1
limit 10` can fail independently of a full-column insert. Please keep at least
one deterministic direct `s3()` assertion over a nested value, for example a
scalar extraction or `size(...)`/casted expression ordered by `k1`, instead of
reducing the direct check to `k1` only.
--
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]
