Csaba Ringhofer has posted comments on this change. ( http://gerrit.cloudera.org:8080/24889 )
Change subject: IMPALA-15106: Support missing types with theta sketches ...................................................................... Patch Set 1: (3 comments) Looks great! I think that it is important though to explore the interoperability landscape a bit, so check some other engines like Hive to avoid issues like the different string hashing in the past. http://gerrit.cloudera.org:8080/#/c/24889/1//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/24889/1//COMMIT_MSG@20 PS1, Line 20: TINYINT | yes | yes | already done : SMALLINT | yes | no | needs FE wiring only It could be noted that these are not supported in Iceberg, as I assume that the main goal of this change is to write theta sketches in Iceberg puffin files. The Iceberg spec defines exactly how to hash its types, but I am not sure if non-Iceberg types are well defined. http://gerrit.cloudera.org:8080/#/c/24889/1//COMMIT_MSG@47 PS1, Line 47: Tests: There should be some test that checks the exact output compared to some reference. For example create a sketch for a table also in Hive and Spark, and check the byte arrays generated. http://gerrit.cloudera.org:8080/#/c/24889/1/be/src/exprs/aggregate-functions-test.cc File be/src/exprs/aggregate-functions-test.cc: http://gerrit.cloudera.org:8080/#/c/24889/1/be/src/exprs/aggregate-functions-test.cc@209 PS1, Line 209: // Three distinct timestamps (0 s, 1 s, 2 s since epoch). It would be nice to also have a negative interval. -- To view, visit http://gerrit.cloudera.org:8080/24889 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I549a51ba96d0b9be2052a9350efe2096c1755816 Gerrit-Change-Number: 24889 Gerrit-PatchSet: 1 Gerrit-Owner: Mihaly Szjatinya <[email protected]> Gerrit-Reviewer: Csaba Ringhofer <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Peter Rozsa <[email protected]> Gerrit-Comment-Date: Thu, 24 Sep 2026 06:38:31 +0000 Gerrit-HasComments: Yes
