impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Armstrong (Code Review)" <>
Date Thu, 23 Jul 2020 06:27:51 GMT
Tim Armstrong has posted comments on this change. ( )


Patch Set 8:

Commit Message:

note to self: need to focus on tests
File testdata/workloads/functional-planner/queries/PlannerTest/setoperation-rewrite.test:
PS8, Line 470: |  hash predicates: bigint_col IS NOT DISTINCT FROM functional.alltypestiny.bigint_col,
bool_col IS NOT DISTINCT FROM functional.alltypestiny.bool_col, double_col IS NOT DISTINCT
FROM functional.alltypestiny.double_col, float_col IS NOT DISTINCT FROM functional.alltypestiny.float_col,
id IS NOT DISTINCT FROM, int_col IS NOT DISTINCT FROM functional.alltypestiny.int_col,
month IS NOT DISTINCT FROM functional.alltypestiny.month, smallint_col IS NOT DISTINCT FROM
functional.alltypestiny.smallint_col, timestamp_col IS NOT DISTINCT FROM functional.alltypestiny.timestamp_col,
tinyint_col IS NOT DISTINCT FROM functional.alltypestiny.tinyint_col, year IS NOT DISTINCT
FROM functional.alltypestiny.year, string_col IS NOT DISTINCT FROM functional.alltypestiny.string_col,
date_string_col IS NOT DISTINCT FROM functional.alltypestiny.date_string_col
> Actually, I was not referring to planning time but the execution time.  I h
Yeah it does add overhead - with the regular equality predicates, we don't insert or probe
with rows with null join keys, so the null check is omitted. In general it would be helpful
to have more nullability info since there are a lot of null checks in the compiled code (basically
every SlotRef expr)

To view, visit
To unsubscribe, visit

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I5be46f824217218146ad48b30767af0fc7edbc0f
Gerrit-Change-Number: 16123
Gerrit-PatchSet: 8
Gerrit-Owner: Shant Hovsepian <>
Gerrit-Reviewer: Aman Sinha <>
Gerrit-Reviewer: David Rorke <>
Gerrit-Reviewer: Impala Public Jenkins <>
Gerrit-Reviewer: Shant Hovsepian <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-Comment-Date: Thu, 23 Jul 2020 06:27:51 +0000
Gerrit-HasComments: Yes

  • Unnamed multipart/alternative (inline, 8-Bit, 0 bytes)
View raw message