impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Tauber-Marshall (Code Review)" <>
Subject [Impala-ASF-CR] IMPALA-3524: Don't process spilled partitions with 0 probe rows
Date Thu, 22 Dec 2016 01:37:19 GMT
Thomas Tauber-Marshall has posted comments on this change.

Change subject: IMPALA-3524: Don't process spilled partitions with 0 probe rows

Patch Set 4:

File be/src/exec/

Line 387:       return mem_tracker()->MemLimitExceeded(
> Yeah I think maybe just making a counter like "NumHashTableBuildsSkipped" m
File be/src/exec/

Line 379:     // If there are no probe rows, there's no need to build the hash table, and
> The logic for !NeedToProcessUnmatchedBuildRows() is a bit convoluted - we s

Line 665:   // This will only be called for partitions that are added to 'output_build_partitions_'
> extra space

PS3, Line 666: processed unti
> OutputAllBuild.

PS3, Line 667: here
> must

PS3, Line 666: ch adds one partition that is then processed until
             :   // it is done by the loop in GetN
> Could condense - I don't think the bit about multiple calls is needed to ex

Line 680:       output_unmatched_batch_iter_.reset(
> Could avoid reallocating by assigning *output_unmatched_batch_iter_ = RowBa
I don't think that works, due to the const members in RowBatch::Iterator

Line 705:   return Status::OK();
> Extra blank line here.
File be/src/exec/partitioned-hash-join-node.h:

PS3, Line 289: OutputUnmatchedBuild
> OutputUnmatchedBuild()

PS3, Line 480: put_unmatched_batch_;
> typo

To view, visit
To unsubscribe, visit

Gerrit-MessageType: comment
Gerrit-Change-Id: I175b32dd9031e51218b38c37693ac3e31dfab47b
Gerrit-PatchSet: 4
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Thomas Tauber-Marshall <>
Gerrit-Reviewer: Thomas Tauber-Marshall <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-HasComments: Yes

View raw message